Anthropic holds most of the measured crowns this month. Claude Opus 5.5 (September 22) is #1 on the Intelligence Index at 57.6 and on Arena's WebDev board, Claude Sonnet 5.5 (September 28) is second at 56.0 and free on claude.ai, and Claude Haiku 5.5 (October 7) completes the family at $0.10 / $0.50. Our guide to AI benchmarks explains how that index is built.

Google's Gemini 4 Argon (September 30) scores 52.6, level with GPT-6 Astra, and is first on Arena's text board, and since October 13 it is open to Google AI Pro and Ultra subscribers and the paid API, which puts it second on the chat podium, but it wins no category: EQ-Bench, LiveBench and ARC Prize have not rated it yet. OpenAI competes on price: GPT-6 Sol and Luna halved its API rates, and GPT-6.1 Sol (September 29) scores 51.8 at $2 / $10, the cheapest model in the top tier per index task.

Meta's Muse Wonder (October 12) is the newest frontier entrant: Muse Wonder Full scores 55.8, third behind the two Claude models, but it runs only inside Meta's Muse Code subscription, and Muse Wonder on the API scores 52.4 at Opus 5.5's $4 / $20. Grok 4.7 (September 21) reached the API at an unchanged $2 / $6, and Xiaomi's MiMo-V2.6-Pro is still the highest-scoring open model at 46.3. On October 6 Mistral released Mistral Large 4 and Google shipped Nano Banana 2.1; neither takes a crown, but video changes hands to MiniMax H3. Rankings are updated within 48 hours of any major launch.

Oct 13 Google Gemini 4 Argon opens to Google AI Pro, Ultra and the paid API, and takes second place on the chat podium New

Google released Gemini 4 Argon to the public on October 13, 2026, two weeks after a launch limited to cyber defenders in its Fairwind Program. It is rolling out in the Gemini app and AI Mode in Search for Google AI Pro ($19.99/mo) and Ultra subscribers, on the paid Gemini API as gemini-4-argon, and for Ultra subscribers in Google Antigravity and AI Studio; the free plan and Google AI Plus do not get it. The introductory $2 / $10 per 1M tokens runs until November 30, 2026, then $4 / $20. The model card gives a March 2026 knowledge cutoff and a 1M-token limit shared by input and output. No board moved with the release: Artificial Analysis still scores it 52.6 and Arena still ranks it first on text at 1525, while EQ-Bench, LiveBench and ARC Prize have not rated it. Now that people can use it, it takes second place on the chat podium and a new legal, finance and tax row in the work table. Full detail in our Gemini 4 Argon breakdown.

Oct 12 Meta Muse Wonder Full lands third on the Intelligence Index, but only inside Meta's Muse Code subscription New

Meta released Muse Wonder and Muse Wonder Full on October 12, 2026, the first models of a larger class above Muse Spark, trained at its Prometheus data center. Artificial Analysis scores Muse Wonder Full 55.8 on Intelligence Index v4.3.2 at max effort, third among distinct models behind Claude Opus 5.5 at 57.6 and Claude Sonnet 5.5 at 56.0, and second on GDPval-AA v2.1 at 1857, between Opus 5.5 and Sonnet 5.5. It takes 57.1% on Terminal-Bench 4.0, level with Gemini 4 Argon and behind Sonnet 5.5's 63.6%. Muse Wonder Full is only available to Muse Code subscribers on the $50 and $200 plans. The smaller Muse Wonder is on the Meta Model API and OpenRouter at $4 / $20 per 1M tokens, the same as Opus 5.5, and scores 52.4 at $4.52 an index task. Meta's own chart puts Wonder Full ahead on long-horizon coding, 57.5% on SWE-Marathon against Opus 5.5's 46.3%, but Arena, LiveBench and EQ-Bench have not rated either model yet. Neither takes a crown on this page. Full detail in our Muse Wonder breakdown.

Oct 7 Anthropic Claude Haiku 5.5 scores 43.4 at GPT-6 Luna's price, and every Claude plan can pick it, Free included New

Anthropic released Claude Haiku 5.5 on October 7, 2026, the third and smallest model in the Claude 5.5 family. It costs $0.10 / $0.50 per 1M tokens for prompts up to 100,000 tokens, the same as GPT-6 Luna, and $0.50 / $2.50 above that, with cache reads at $0.01. Anthropic says it runs about 75% cheaper than Claude Haiku 4.5 at $1 / $5, a figure that already allows for a newer tokenizer that counts the same text as about 30% more tokens. Artificial Analysis scores it 43.4 on Intelligence Index v4.3.2 at max effort, up from 17 for Haiku 4.5 and ahead of GLM 5.3 Flash at 41.8, Gemini 3.8 Flash at 40.9 and GPT-6 Luna at 38.1, though an index task costs $0.21 against Luna's $0.07 because it writes far more output. It knows less than those rivals but guesses less: 36% correct on AA-Omniscience with a 40% hallucination rate, against Luna's 77%. LiveBench puts it at 57.2 on Agentic Coding, just above Claude Sonnet 5.5's 56.3, and Arena has rated it only on WebDev, 30th at 1587. It has a 1M-token context, 128K output and, for the first time on a Haiku, an effort setting. Anthropic says Free, Pro, Max, Team and Enterprise users can select it on Claude.ai on the web, iOS and Android, and that Sonnet 5.5 and Opus 5.5 remain the better choices for complex agentic coding. It takes no crown on this page. Full detail in our Claude Haiku 5.5 guide.

Oct 6 Mistral Mistral Large 4 is the strongest model from the US or Europe that promises open weights, but you cannot download it yet New

Mistral released Mistral Large 4, nicknamed Le Chonk, as a public preview API on October 6, 2026. It is a 1.05-trillion-parameter mixture of experts with 49 billion active at launch (Mistral's docs card now says 52 billion), reads text and images, and lists at $1.36 / $4.18 per 1M tokens, though Mistral's own model page shows half that without saying why or for how long. Artificial Analysis scores it 38.4 on Intelligence Index v4.3.2, far above the best US open model, NVIDIA's Nemotron 3 Ultra at 23, but five to eight points behind GLM-5.3, Kimi K3 and MiMo-V2.6-Pro, and it is verbose enough that an index task costs $1.13. Arena has not rated it yet. Mistral says the weights follow by the end of October and has not named a licence. Our Mistral Large 4 breakdown covers the benchmarks and the cyber-defense pitch.

Oct 6 Google Nano Banana 2.1 replaces Nano Banana 2 at half the price per image and lands fifth on Arena text-to-image New

Google released Nano Banana 2.1 on October 6, 2026 as the direct successor to Nano Banana 2 (Gemini 3.1 Flash Image) and deprecated Nano Banana 2 the same day, with an API shutdown on October 29. In the Gemini API an image costs $0.0336 per 1K image against $0.067 before, but input tokens triple to $1.50 per 1M and the cheap 512px size is gone. Arena ranked it fifth on text-to-image at 1328.1, in the same rank band as MAI-Image-2.6 and Grok Imagine Image 2.0, and sixth on image editing at 1428.1, the highest of any Google model on both boards, while ChatGPT Images 2.5 still holds the top two places on each. Artificial Analysis has not rated it yet. It is in the Gemini app, AI Mode in Google Search, Google AI Studio, Flow and the Gemini API. Our Nano Banana 2.1 guide covers what changed and the real cost.

Sep 30 Google Gemini 4 Argon tops Arena's text board and matches GPT-6 Astra on the Intelligence Index, but only cyber defenders can use it New

Google announced Gemini 4 Argon on September 30, 2026, the first model of the Gemini 4 generation and its first flagship since Gemini 3.1 Pro in February. It is a reasoning model aimed at long, complex work, coding, legal and financial research and cybersecurity defense, and it can write up to 1M output tokens in one response, up from 64K. Artificial Analysis scores it 52.6 on Intelligence Index v4.3.2 at high effort, level with GPT-6 Astra at 52.7 and fifth among distinct models behind Claude Opus 5.5, Claude Sonnet 5.5 and Claude Fable 5.1, at $1.99 an index task on the introductory price. Its standout result there is reliability: a 15% hallucination rate on AA-Omniscience, the lowest Artificial Analysis has measured among leading models, though it answers only 50% of the questions correctly against 67% for Fable 5.1. It also leads AutomationBench-AA at 77.5%, ahead of Claude Sonnet 5.5's 71.3%, but trails on Terminal-Bench 4.0 at 57.1% against Sonnet 5.5's 63.6%. Within a day Arena ranked it first on text at 1524.8 and on creative writing and math, but only eighth on WebDev at 1679. On Google's own table it leads DeepSWE v1.1 at 77.9% and the Vals Index, and loses Terminal-Bench 4.0 to Claude Opus 5.5 and FrontierSWE v2 to GPT-6 Astra. Access is the catch: Argon went only to vetted cyber defenders in Google's Fairwind Program, with paid API customers and Google AI Ultra subscribers next and no date given. The introductory price is $2 / $10 per 1M tokens, with cached input 95% off, against a standard $4 / $20. No crown on this page moves until people can use it and the other boards rate it. Full detail in our Gemini 4 Argon breakdown.

Sep 29 OpenAI GPT-6.1 Sol reaches fifth on the Intelligence Index at $2 / $10, the cheapest per task in the top tier New

OpenAI released GPT-6.1 Sol at DevDay on September 29, 2026, a week after GPT-6 Sol, at the same $2 / $10 per 1M tokens with cached input halved to $0.10. OpenAI says it nearly matches GPT-6 Astra on agentic coding, computer use and professional work at a fifth of Astra's token prices, and the independent numbers put it close rather than level. Artificial Analysis scores it 51.8 on Intelligence Index v4.3.2 at max effort, fifth among distinct models behind Claude Opus 5.5 at 57.6, Claude Sonnet 5.5 at 56.0, Claude Fable 5.1 at 53.4 and Astra at 52.7, against 47.5 for GPT-6 Sol, and an index task costs $0.72, against $3.26 for Astra and $5.98 for Opus 5.5. LiveBench puts it second on Reasoning at 92.63, 0.02 behind Astra, third on Mathematics at 96.83 and second on Language at 90.13, but well outside the top ten on Coding at 80.7. It takes 56.1% on Terminal-Bench 4.0 against GPT-6 Sol's 43.9% and scores 1575 on GDPval-AA v2.1, above Astra's 1542. Arena, EQ-Bench and ARC Prize have not rated it yet. It is in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu, not in Chat, and every GPT-6 tier bills 2x input and 1.5x output on the whole request once a prompt passes 272K tokens. OpenAI shipped no GPT-6.1 Astra beside it; the Wall Street Journal reported that its October release was cancelled after internal safety tests. The same event launched OpenAI dots, always-on agents that run on GPT-6 Astra, and a $500 ChatGPT Pro plan. Since then Arena has put it third on WebDev at 1759, ARC Prize has scored it 94.2% on ARC-AGI-2, second only to Astra, and Gemini 4 Argon has pushed it to sixth on the index.

Sep 28 Anthropic Claude Sonnet 5.5 lands second on the Intelligence Index at 56.0, at Sonnet's $2 / $10 New

Anthropic released Claude Sonnet 5.5 on September 28, 2026, the second model in the Claude 5.5 family, at the same $2 / $10 per 1M tokens as Sonnet 5, with cache reads at $0.20. Anthropic says it runs more than 30% faster and costs up to 30% less per task, and the independent numbers back the capability claim: Artificial Analysis measures it at 56.0 on Intelligence Index v4.3.2 at max effort, second only to Claude Opus 5.5 at 57.6 and ahead of Claude Fable 5.1 at 53.4 and GPT-6 Astra at 52.7, against 38.2 for Sonnet 5. It takes Artificial Analysis's Terminal-Bench 4.0 run at 63.6% against Opus 5.5's 59.6%, LiveBench Coding at 91.4 against 89.3 and AutomationBench-AA at 71.3% against 69.5%, and it sits two Elo points behind Opus 5.5 on GDPval-AA v2.1, 1844 to 1846. The catch is effort. At max, Sonnet 5.5 spent 410 million tokens on the index and a task costs $7.60, against $5.98 for Opus 5.5 and $5.09 for Sonnet 5. At xhigh it scores 51.9 for $2.74 a task, close to GPT-6 Astra's 52.7 for less than Astra's $3.26, and at high 46.7 for $1.08. The Claude apps run it at Medium effort by default and the API at High. It is weaker where Opus 5.5 is strongest: 56.3 on LiveBench Agentic Coding against 71.7, and 32.3 on the hallucination-penalised AA-Omniscience index against 46.4. Arena has not ranked it yet. Anthropic says anyone can chat with Sonnet 5.5 on claude.ai, and higher-risk cybersecurity requests fall back to Sonnet 5, which is now a legacy model. Arena has since ranked it fifth on WebDev, at 1709, and Anthropic halved its cache-read price to $0.10 on October 7.

Sep 22 OpenAI GPT-6 Sol and GPT-6 Luna halve OpenAI's API prices, and land everywhere except ChatGPT's chat window New

OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, 2026, about ninety minutes after Anthropic shipped Claude Opus 5.5, and the launch is an argument about cost per task rather than about the top of the board. Sol drops from $4 / $20 to $2 / $10 per 1M tokens and Luna from $0.20 / $1.20 to $0.10 / $0.50, with cached input reads discounted 90%, to $0.20 and $0.01. OpenAI labels both cuts 50%, but Luna's output falls 58.3%, and the comparison is against GPT-5.6 promotional pricing rather than list. Two spokespeople and OpenAI's Tibo Sottiaux all say the new rates are permanent. Where you can use them is the part every summary flattened: the models are in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu, Free and Go users reach Luna in the desktop app, and OpenAI states plainly that they are "not yet available in Chat". Independently, Artificial Analysis scores Sol 47.5 on Intelligence Index v4.3.2 at max effort against GPT-5.6 Sol's 47.0, at $1.06 an index task against $1.99, with Terminal-Bench 4.0 at 43.9% against 39.9% and the hallucination rate down from 92% to 60%. Luna is level with GPT-5.6 Luna at 37.3 but runs an index task for $0.07 against $0.18, which makes it the cheapest model per index task on this page. ARC Prize has scored Luna, 59.3% on ARC-AGI-2 against the 5.6 build's 59.5%, and has published nothing for Sol. OpenAI's own evaluations are all cost arguments: on AutomationBench 1.0.6 Sol at xhigh effort takes 33.2% at $0.27 a task, and the closest rival row is Claude Fable 5.1 with an Opus 5 fallback at 31.4%, so the real lead is 1.8 points rather than the 6.3 the Claude Opus 5 row implies. OpenAI says Sol makes about half as many mistakes as GPT-5.6 Sol, "approaching Astra-level reliability at much lower cost", and repeats that Astra is still its best model across the board. Nothing here moves a crown: Sol and Luna are cheaper, not better. There is no GPT-6 Terra, and OpenAI has not said whether one is planned. Full detail in our GPT-6 Sol and Luna breakdown.

Sep 22 Anthropic Claude Opus 5.5 takes the Intelligence Index at 57.6 and undercuts Opus 5 on price New

Anthropic released Claude Opus 5.5 on September 22, 2026, the first model in a new Claude 5.5 family, and it is the largest single-day move this page has recorded. Artificial Analysis measures it at 57.6 on Intelligence Index v4.3.2, 4.3 points above Claude Fable 5.1 and the highest score it has published for any model, and it takes GDPval-AA v2.1 at 1846 against Fable 5.1's 1735, AA-Briefcase v1.1 at 1822 against 1678, Humanity's Last Exam at 61.4% against 59.1% and Artificial Analysis's own Terminal-Bench 4.0 run at 59.6% against GPT-6 Astra's 59.1% with it. Pricing goes down rather than up: $4 / $20 per 1M tokens against Opus 5's $5 / $25, cache reads 60% cheaper at $0.20 and cache writes at $5, with Anthropic's own testing putting a typical workload 40% cheaper than on Opus 5 and output more than 30% faster. A fast mode in Claude Code and the Claude Platform doubles the price for up to 2.5x the speed. On Anthropic's own harness the coding gap is wider still, 66.4% on Terminal-Bench 4.0 against the 57.9% OpenAI reports for Astra, plus 54.4% on FrontierCode v1.1 and 57.8% on CursorBench 4.0. Two things it does not win: Zapier's AutomationBench, where Astra scores 41.4% to its 40.0%, and Terminal-Bench-Science 0.1, where Astra scores 64.6% to its 58.7%. Read the margins with the caveat Anthropic prints itself, that the model "performs at the level of Claude Fable 5.1 on most work" and that "benchmark margins have become a less reliable guide to real-world differences". It is generally available on the Claude Platform as claude-opus-5-5 and on AWS, Google Cloud and Microsoft Azure, five-hour limits rise on Pro, Max and Team, and Claude Sonnet 5.5 and Claude Haiku 5.5 follow in the coming weeks. Because it matches Claude Mythos 5.1 in biology and cybersecurity it ships with Fable 5.1's safeguards and a new Life Sciences Verification Program for vetted labs. No Arena board has rated it yet. Arena has since ranked it first on WebDev and second on text, and Claude Sonnet 5.5 and Claude Haiku 5.5 followed on September 28 and October 7. Full detail in our Claude Opus 5.5 breakdown.

Sep 21 SpaceXAI Grok 4.7 arrives on a larger base model at an unchanged $2 / $6, and costs 47% more per task New

SpaceXAI released Grok 4.7 on September 21, 2026 as its most capable model for coding and knowledge work. It uses a new and larger base model than Grok 4.6, with a longer reinforcement learning run weighted toward tasks that take many hours, and it was trained to understand the Grok Bot harness natively. The company publishes no parameter count. On its own figures it takes 46.3% on CursorBench 4.0 against Grok 4.6's 40.4% and GPT-5.6 Sol's 41.7%, 71.0% on DeepSWE v1.1 at high effort, 64.0% on EEBench, 38.0% on Terminal-Bench 4.0, 19.6% on the Harvey legal agent benchmark and 56.7% on HealthBench Professional. Independently, Artificial Analysis puts it at 46.3 on Intelligence Index v4.3.2 at high effort against Grok 4.6's 44.3, and at 1695 on GDPval-AA v2.1 and 1657 on AA-Briefcase v1.1, both above GPT-6 Astra's 1542 and 1569. The price does not move, $2 / $6 per 1M tokens with the whole request re-billed at $4 / $12 above 200,000 prompt tokens, and a fast variant runs at twice the output speed for twice the price. What does move is the cost per task: $2.73 against Grok 4.6's $1.86, because 4.7 emits roughly twice the output tokens getting to the same place, and xhigh effort buys no extra index points on either generation. On safety SpaceXAI reports the strongest refusal and jailbreak resistance it has tested, 62.4% on LatchBio's biosafety benchmark and 3.3% of risky dual-use prompts allowed through on its own HackerBench v0.3. Availability is developer-side: the Grok API, Cursor, Grok Build, third-party coding harnesses and cloud routers. The announcement said nothing about the consumer Grok app, which moved to Grok 4.7 on October 3. Full detail in our Grok 4.7 breakdown.

Sep 21 Xiaomi Xiaomi's MiMo-V2.6-Pro is the highest-scoring open model ever measured, at $0.13 an index task under MIT New

Xiaomi published MiMo-V2.6-Pro on September 21, 2026 under plain MIT, with the weights on Hugging Face at XiaomiMiMo/MiMo-V2.6-Pro-RL on the day. It is a 1.02-trillion-parameter mixture of experts with 42 billion active and a 1M-token context. Artificial Analysis measures it, rather than estimating it, at 46.3 on Intelligence Index v4.3.2, which is the highest open score it has ever published: above GLM-5.3 at 44.8, above Kimi K3 at 43.6 and level with SpaceXAI's brand-new Grok 4.7. The cost is the other half of the story. It runs an Intelligence Index task for $0.13 at $0.435 / $0.87 per 1M tokens, which is cheaper per task than any other open model on this page, roughly half what GLM 5.3 Flash costs for four and a half points more. It also takes 34.8% on Terminal-Bench 4.0, 1673 on GDPval-AA v2.1, 49.4% on Humanity's Last Exam and 86.3% on AA-LCR, which puts it above Kimi K3 on three of those four. Two limits are worth stating plainly. No Arena board has rated it, so there is no human-preference evidence for it at all, and it was published four days ago, so it has no independent track record beyond Artificial Analysis. And 1.02 trillion parameters is a cluster, not a workstation, which is why GLM 5.3 Flash stays our recommendation for teams that actually want to host their own. It takes the open-weight crown on this page on measured score and cost. Arena has since rated it, 23rd on WebDev and 25th on text.

Sep 20 Alibaba Qwen-Image-2.1 is the best image model you can download, but its licence allows research use only New

Alibaba's Qwen team released Qwen-Image-2.1 on September 20, 2026, on Hugging Face, ModelScope and GitHub at once, and it is now the best image model you can download. It is one model for both generating and editing, with a 7B-parameter generator, down from about 20 billion in the original Qwen-Image. It makes transparent PNGs natively and edits with up to 10 reference images. On the independent boards it is the highest-ranked open-weight model on all four: 17th on Arena image editing at 1366.4, 18th on Arena text-to-image at 1227.6 and 18th on both Artificial Analysis boards. The catch is the licence. The 2.1 weights ship under the Qwen Research License for "research or evaluation purposes only", a change from the Apache 2.0 terms of earlier Qwen-Image releases, so it is not a free commercial option. For commercial work the older Apache 2.0 releases are the safe choice, though they rank far lower, Qwen-Image-2512 43rd on Arena text-to-image. Alibaba's hosted Qwen-Image-3.0-Pro, launched in July, ranks higher at 12th on Arena text-to-image, but it has no weights and costs $0.04 to $0.075 per image through the API. Full detail in our Qwen image generator guide.

Sep 15 TypeSafe TypeSafe leaves stealth with Jev, a model that returns typed decisions instead of text, at $0.042 per 1M tokens New

TypeSafe AI came out of two years of stealth on September 15, 2026 with $40 million in seed funding led by DCVC and a model that does not generate text at all. Jev is the first of what the company calls System One models: you send a block of state plus a fixed set of typed questions, and it answers all of them in one parallel pass, returning a choice, a score or a yes-no probability instead of a sentence. Because the answer space is defined before the call it cannot return a malformed result, which is where the launch post's line that Jev can't hallucinate comes from. It can still pick the wrong option, and TypeSafe's own documentation says calibration does not guarantee that an individual answer is correct. Pricing is $42 per billion input tokens, which is $0.042 per 1M, with output free, on a 64k context and text input only. The performance claims need care, because the company published four different ones in three days: from the press release's under 100 milliseconds and up to 100x faster to the homepage's 193.6x faster and 444.6x cheaper, each measured against a different rival, with its own footnote calling the big numbers the higher end of real world gains. Its workflow eval scores against the average of GPT-6 Astra and Fable 5.1 rather than ground truth, and it publishes no public benchmark scores at all by choice. Two independent early-access tests on September 18 measured roughly 6x cheaper and 8x faster against a cheap model with reasoning off, and the calibration held up. Nothing here moves a pick on this page: Jev has no consumer surface and is waitlisted early access. Full detail in our System One models explainer.

Pending Google Gemini 3.5 Pro, still unreleased, and Gemini 4 Argon has overtaken it Delayed

Gemini 3.5 Pro is still unreleased. Google announced it at I/O on May 19 alongside Gemini 3.5 Flash, but only Flash shipped, and Bloomberg reported on July 16, 2026 that it was months behind schedule. Google DeepMind's Koray Kavukcuoglu said on September 24 that the company "took a little bit of a step back" to focus on its Flash models and on Gemini 4, and on September 30 Google announced Gemini 4 Argon as its next flagship instead. Google has not said whether 3.5 Pro will ship at all. Use Gemini 3.8 Flash in the meantime if you are working through the API, or Gemini 3.6 Flash in the Gemini app, where since October 9, 2026 it needs Google AI Plus or above.

Category pick, October 2026

Best AI for Writing

Claude Fable 5.1

Best writing overall

→

Kimi K3

Creative fiction

→

Claude Sonnet 5.5

Free everyday

The best AI for writing is Claude Fable 5.1, and the case for it is consistency rather than a single first place.

It is inside the top ten of all three prose boards: second on EQ-Bench Creative Writing v3 at 2162.0, third on LiveBench Language at 89.5 and ninth on Arena's creative-writing board. No other model places that well across all three. Gemini 4 Argon now leads Arena's creative-writing and text boards, at 1519.2 and 1525.4, and Google AI Pro subscribers can use it since October 13, but neither EQ-Bench nor LiveBench has rated it yet.

Claude Opus 5.5 is second on Arena's creative-writing board at 1517 but seventh on EQ-Bench and tenth on LiveBench Language; GPT-6 Astra leads EQ-Bench at 2173.3 but sits 42nd on Arena's creative-writing board; Claude Fable 5 tops LiveBench Language at 90.7 and is third on Arena's creative-writing board, but eleventh on EQ-Bench. If your writing is a work deliverable rather than prose, the GDPval-AA v2.1 professional-deliverables board is a different question, and Claude Opus 5.5 leads it at 1866, ahead of Meta's new Muse Wonder Full at 1857, Claude Sonnet 5.5 at 1838 and Claude Fable 5.1 at 1756.

Claude Sonnet 5.5 is the free-tier value pick, since anyone can use it on claude.ai, though it scores 83.4 on LiveBench Language, sits only 24th on Arena's creative-writing board and has no EQ-Bench rating yet. Claude Fable 5 is the pick if you weight Arena's human votes above the harness scores; it costs $10 / $50 per 1M tokens and is permanently included in Claude Max and Team Premium at roughly 50% of regular usage limits. Translation is a separate question with a separate winner, which we work through in our guide to the best AI for translation.

Model	Best For	Strength	Weakness	Price (per 1M tokens)
Claude Fable 5.1	Best writing overall	#2 EQ-Bench Creative Writing (2162.0), #3 LiveBench Language (89.5) and #9 Arena creative writing, the best combined placing of any model	Leads none of the three outright; Gemini 4 Argon now leads Arena's creative-writing board, with Claude Opus 5.5 second	$10 / $50
Claude Fable 5	Strong on the human-vote boards	#1 LiveBench Language (90.7), #3 Arena creative writing (1501.5), #4 Arena text (1503.7)	#11 EQ-Bench; priciest option here	$10 / $50
Kimi K3	Creative fiction and voice	#5 EQ-Bench Creative Writing (2082.3)	Only #28 on Arena creative writing	$3 / $15
Claude Sonnet 5.5	Free everyday writing	Free on claude.ai, 1M context; #3 GDPval-AA v2.1 (1838), behind Opus 5.5 and Muse Wonder Full	83.4 LiveBench Language; #24 Arena creative writing; not yet rated on EQ-Bench	$2 / $10
Claude Opus 5.5	Arena's top creative writer you can use	#2 Arena creative writing (1517) and #2 Arena text (1507); #1 GDPval-AA v2.1 (1866)	#7 EQ-Bench and #10 LiveBench Language	$4 / $20
Gemini 4 Argon	Top-voted prose on Arena	#1 Arena creative writing (1519.2) and #1 Arena text (1525.4)	Unrated on EQ-Bench and LiveBench	$19.99/mo Google AI Pro; $2 / $10 API intro to Nov 30
GPT-5.5	Fact-anchored business writing	Documented factual-reliability gains over GPT-5.4	Reasoning tiers now marked deprecated	$5 / $30
Gemini 3.8 Flash	Bulk drafts at scale	Intelligence Index 40.9, 308.4 tok/s output, 1M context	Agent-tuned rather than prose-tuned; intro price doubles January 1, 2027	$0.75 / $3.75

Runner-up and alternatives: Claude Fable 5 is the runner-up and the pick if you weight Arena's human votes, Claude Opus 5.5 is the highest-placed model you can use on Arena's creative-writing board and the leader for professional deliverables, Kimi K3 is the pick for distinctive fiction though it now sits fifth on EQ-Bench, Claude Sonnet 5.5 is the value pick and the one to use if you are not paying, and Gemini 3.8 Flash is the pick for bulk drafting. Gemini 4 Argon, on Google AI Pro since October 13, is the one to watch once EQ-Bench and LiveBench rate it.

What changed this month

The writing crown stays with Claude Fable 5.1, but its margin on Arena narrowed. Google's Gemini 4 Argon, announced September 30, went straight to first on Arena's creative-writing board at 1519.2, pushing Claude Opus 5.5 to second and Claude Fable 5 to third, and Fable 5.1 slipped from seventh to tenth and has since recovered to ninth, so this card's claim is now the top ten of all three prose boards rather than the top seven. Argon cannot take the crown yet: since October 13 Google AI Pro and Ultra subscribers can use it, but EQ-Bench and LiveBench have still not rated it. In September, Claude Opus 5.5 and Claude Sonnet 5.5 arrived, Sonnet 5.5 replaced Sonnet 5 as the free pick, and GPT-6.1 Sol entered LiveBench Language second, pushing Fable 5.1 to third there.

Category pick, October 2026

Best AI for Chat & Daily Assistant

GPT-6

ChatGPT's default

→

Gemini 4 Argon

#1 on Arena text

→

Claude Opus 5.5

#1 on the index

The best AI for everyday chat is GPT-6, and the honest reason is reach rather than board position. Since October 7 it is what ChatGPT answers with by default: GPT-6 Sol on Plus, Pro, Business and Enterprise, and GPT-6 Luna on Free and Go, rolling out from October 8. That makes it the assistant most people can actually open.

On raw human preference it is far from the lead. GPT-6 Sol sits #67 on Arena's text leaderboard at 1455.9, below its predecessor GPT-5.6 Sol at #20, while Gemini 4 Argon leads at 1525.4 and Claude Opus 5.5 is second at 1507.1. Artificial Analysis scores GPT-6 Sol 47.6 on Intelligence Index v4.3.2 at maximum effort and GPT-6 Luna 38.1. In the API, Sol costs $2 / $10 and Luna $0.10 / $0.50 per 1M tokens.

Model	Best For	Strength	Weakness	Price
GPT-6	Everyday chat, ChatGPT's default	The assistant most people can open: Sol on paid plans, Luna on Free and Go	GPT-6 Sol only #67 on Arena text	Free / $20/mo Plus; API $0.10 / $0.50 (Luna) to $2 / $10 (Sol)
Gemini 4 Argon	Top-voted conversation	#1 on Arena text (1525.4); 15% hallucination rate, the lowest among leading models	Arena rating still preliminary; Google AI Pro or above in the Gemini app	$19.99/mo Google AI Pro; $2 / $10 API intro to Nov 30
Claude Opus 5.5	Highest-scoring conversation	#1 on the Intelligence Index (57.6) and #2 on Arena text (1507.1)	Not on the Free plan; usage-limited on Pro	$20/mo Pro, $4 / $20 API
Claude Sonnet 5.5	Free Claude chat	Free on claude.ai; #2 on the Intelligence Index (56.0), behind only Claude Opus 5.5	Only #32 on Arena text; runs at Medium effort by default in the apps	Free / $20/mo Pro, $2 / $10 API
GPT-5.5 Instant	Hallucination-sensitive daily work	52.5% fewer hallucinated claims vs 5.3 Instant	Reasoning tiers now marked deprecated	$20/mo Plus; API $5 / $30
Gemini 3.6 Flash	Fast, cheap, multimodal	1M context, #23 on Arena text	Needs Google AI Plus in the Gemini app since October 9; 3.8 Flash outscores it at the same price	AI Plus / $0.75 / $3.75 API
Fello AI	Many leading models, one app	ChatGPT + Claude + Gemini + Grok + DeepSeek and more on Mac, iPhone and iPad	Routed via app, not direct	From $11.99/mo

Runner-up and alternatives: Gemini 4 Argon is the runner-up, first on Arena's text board and on Google AI Pro since October 13, Claude Opus 5.5 is the top scorer on Artificial Analysis's index, Claude Sonnet 5.5 is the pick for free Claude chat, Gemini 3.6 Flash is the runner-up for fast and cheap (Google AI Plus in the Gemini app since October 9), and Grok 4.7 is the niche pick for live-news days. Fello AI puts ChatGPT, Claude, Gemini and Grok in one window, so a second opinion is one click away.

What changed this month

The chat pick moves from GPT-5.6 to GPT-6, for the same reason GPT-5.6 held it: on October 7 OpenAI made GPT-6 ChatGPT's default, Sol on paid plans and Luna on Free and Go from October 8. The vote-based boards rate it below its predecessor, GPT-6 Sol #67 on Arena text against GPT-5.6 Sol's #20, which is why this pick rests on reach. Since October 13 Gemini 4 Argon is in the Gemini app for Google AI Pro and Ultra subscribers, so the leader of Arena's text board, at 1525.4, takes second place on the podium; Claude Opus 5.5 moves to third and Claude Sonnet 5.5 stays in the table as the free Claude pick.

Category pick, October 2026

Best AI for Images

ChatGPT Images 2.5

Text in images

→

MAI-Image-2.6

#3 AA image editing

→

Grok Imagine Image 2.0

Fewest restrictions

The best AI for image generation is ChatGPT Images 2.5, which OpenAI made the default on every ChatGPT tier on September 8 and which leads all four image boards we track.

Its two models finish first and second on each: GPT-Image-2.5 Sunburst, built for precision, leads Arena text-to-image at 1424.6 and Arena image editing at 1523.6, and Artificial Analysis's AA-Image-T2I v2.0 at 1197 and AA-Image-Editing v2.0 at 1182, with the faster GPT-Image-2.5 Flare second on all four. Read the Sunburst-to-Flare gap on Arena text-to-image as provisional, because each rests on about 17,000 votes against GPT Image 2's 93,000. MAI-Image-2.6 is the runner-up, third on Artificial Analysis's editing board and fifth on Arena's text-to-image board at 1332.1, a statistical tie with Grok Imagine Image 2.0 just above it.

Third is SpaceXAI's Grok Imagine Image 2.0, released in August and available in Grok Imagine and the xAI API: it is fourth on Artificial Analysis's text-to-image board, on Arena image editing and now on Arena text-to-image, ahead of Meta's Muse Image on three of the four boards. Google's Nano Banana 2.1, released on October 6, is now the highest-ranked Google model: sixth on Arena text-to-image at 1327.8, in the same rank band as MAI-Image-2.6 and Grok Imagine Image 2.0, and sixth on Arena editing, though Artificial Analysis has not rated it yet.

The best image model you can download is Alibaba's Qwen-Image-2.1, released on September 20 and now the highest-ranked open-weight model on all four boards, 18th on Arena image editing and on both Artificial Analysis boards and 19th on Arena text-to-image, but its licence allows research and evaluation only, so it is not a free commercial option. Our full breakdown of what shipped is in ChatGPT Images 2.5.

Model	Best For	Strength	Weakness	Price
ChatGPT Images 2.5	Images with readable text	#1 and #2 on all four boards: Arena text-to-image (1424.6 / 1397.6), Arena editing (1523.6 / 1481.0), Artificial Analysis text-to-image (1197 / 1191) and editing (1182 / 1162)	Its Arena text-to-image lead rests on about 17,000 votes per model	Default on every ChatGPT tier
MAI-Image-2.6	Editing an image you already have	#3 on Artificial Analysis image editing (1137) and #5 on Arena text-to-image (1332.1)	Public preview, not in Copilot or Bing yet; lost the Artificial Analysis editing lead to ChatGPT Images 2.5	MAI Playground / Microsoft Foundry
Grok Imagine Image 2.0	Fewest restrictions, Spicy Mode	#4 on Artificial Analysis text-to-image (1155) and #4 on Arena image editing (1439.1); #4 on Arena text-to-image (1333.5)	Only #10 on Artificial Analysis image editing	$30/mo SuperGrok; xAI API
Nano Banana 2.1	Google's everyday image model	#6 Arena text-to-image (1327.8) and #6 Arena image editing (1428.1), the highest of any Google model; half Nano Banana 2's API price per image	Not yet rated by Artificial Analysis; input tokens cost three times as much as on Nano Banana 2	Gemini app / AI Studio; $0.0336 per 1K image
Seedream 5.0 Pro	Multilingual text + region editing	10+ languages incl. Arabic RTL, lasso and layer editing	Around tenth on the independent boards; copyright cloud	BytePlus / Magnific
Midjourney v8	Stylized art, illustration	Aesthetic baseline most artists prefer	Weaker on text in image	$10-$120/mo
Qwen-Image-2.1	Best image model you can download	Highest-ranked open-weight model on all four boards: #18 Arena image editing (1368.7), #19 Arena text-to-image (1222.9) and #18 on both Artificial Analysis boards; 7B generator, transparent PNGs, up to 10 reference images	Qwen Research License, research or evaluation only, no commercial use	Free download (Hugging Face)
Qwen-Image-3.0-Pro	Dense, text-heavy layouts	#13 Arena text-to-image (1255.2) and #14 Artificial Analysis text-to-image (1090); prompts up to 4.5k tokens, text in 12 languages	Hosted only, no weights; not rated on Arena image editing	Qwen Chat; $0.04-$0.075 per image in the API
Muse Image	Image editing, Meta ecosystem	#6 on Artificial Analysis image editing (1118) and #7 on its text-to-image board	Off the podium; #9 on Arena text-to-image and #8 on Arena editing	Meta AI app

Runner-up and alternatives: MAI-Image-2.6 is the runner-up overall and the pick for editing an image you already have, and Grok Imagine Image 2.0 is third and still the only frontier model that allows Spicy Mode adult content. Muse Image is the pick inside Meta's apps, Nano Banana 2.1, which replaced Nano Banana 2 on October 6, is the pick inside Google's apps, and Reve 2.1 remains the pick for layout, typography and native 4K, seventh on Arena text-to-image. If you want to run a model on your own hardware, Qwen-Image-2.1 scores best of any open model, but only for research and evaluation; for commercial work the older Apache 2.0 Qwen-Image releases are the safe choice, though they rank far lower, Qwen-Image-2512 44th on Arena text-to-image.

What changed this month

ChatGPT Images 2.5 keeps the image crown and still holds first and second on all four boards we track. Since the last update Google released Nano Banana 2.1 on October 6, and Arena ranked it fifth on text-to-image and sixth on editing, the highest of any Google model, inside the same rank band as MAI-Image-2.6 and Grok Imagine Image 2.0. It takes Nano Banana 2's table row rather than a podium place: it now trails both MAI-Image-2.6 and Grok Imagine Image 2.0 on both Arena boards, and Artificial Analysis has not rated it yet.

Grok Imagine Image 2.0 has since climbed to fourth on Arena text-to-image, 1.4 points above MAI-Image-2.6 in fifth; Nano Banana 2.1 is now sixth and Reve 2.1 seventh. Earlier this month Grok Imagine Image 2.0 took third place from Muse Image, and Artificial Analysis moved both its image boards to v2.0. In September, Images 2.5 replaced GPT Image 2 at the top on September 8.

Category pick, October 2026

Best AI for Video

MiniMax H3

#1 Arena image-to-video

→

Gemini Omni 1.1 Flash

#1 Arena text-to-video

→

Wan 3.0

#1 AA text-to-video

The best AI for video generation is MiniMax H3, which MiniMax released on July 31 with open weights, and the case for it is that it places near the top everywhere rather than winning one board outright.

It leads Arena's image-to-video board at 1495.4, ahead of Gemini Omni 1.1 Flash at 1488.5 and Alibaba's Wan 3.0 at 1479.8; on Artificial Analysis's image-to-video board it is second at 1181, behind only MiniMax H3 Max, a model Fal built on top of it; and it is fourth on Artificial Analysis's text-to-video board at 1137. Its weak spot is Arena's text-to-video board, where it is eighth at 1460.3.

It makes 2K clips of 4 to 15 seconds with native stereo audio, from $0.13 a second at 2K, and the weights are on Hugging Face under MiniMax's own community licence, without the 2K upscaler and with an application required in the US, EU, UK and South Korea. Gemini Omni 1.1 Flash, our pick until this update, still leads Arena's text-to-video board at 1516.2, in a statistical tie with Gemini Omni Flash at 1512.9, and it is the pick for long takes: scene extension in 10-second steps to 40 seconds, first and last frame control and 4K output, from about $0.03 a second at 360p to $0.30 at 4K.

But Artificial Analysis has now rated it, and it comes eighth on text-to-video at 1113 with no image-to-video rating at all. Wan 3.0 leads Artificial Analysis's text-to-video board at 1156 and is third on Arena's image-to-video board, but only sixth on Arena's text-to-video board. If you need longer takes inside Google's tooling, Veo 3.1 still runs in the Gemini app, AI Studio and Vertex AI with native audio and 1080p output.

Model	Best For	Strength	Weakness	Price
MiniMax H3	Best AI video overall	#1 Arena image-to-video (1495.4), #2 Artificial Analysis image-to-video (1181), #4 Artificial Analysis text-to-video (1137); 2K clips of 4-15s with native stereo audio	#8 on Arena text-to-video (1460); the open weights exclude the 2K upscaler and need an application in the US, EU, UK and South Korea	$0.13/sec at 2K
Gemini Omni 1.1 Flash	Long takes and 4K	#1 Arena text-to-video (1516), #2 Arena image-to-video (1488.5); 40-second scene extension, 4K output	Only #8 on Artificial Analysis text-to-video (1113); no Artificial Analysis image-to-video rating	$0.03-$0.30/sec; Gemini API / AI Studio / Flow
Gemini Omni Flash	Cheaper 10-second clips	#2 Arena text-to-video (1513), #3 Artificial Analysis image-to-video (1178); conversational editing	Caps at 10-second generations	~$0.10/sec; Gemini app / AI Studio
Wan 3.0	Video from documents and slides	#1 Artificial Analysis text-to-video (1156), #3 Arena image-to-video (1479.8); 2-30s, up to 1080p, Omni-Reference input	API-only, no published weights; #6 on Arena text-to-video	$0.05-$0.20/sec
Dreamina Seedance 2.5	ByteDance challenger	#3 Artificial Analysis text-to-video (1143), #4 Arena image-to-video (1477.4)	#7 Arena text-to-video (1474); ByteDance ecosystem, limited Western access	Dreamina / BytePlus
Muse Video	Meta ecosystem video	#9 on Arena text-to-video at 1456	Newest of the group, thin tooling	Meta AI app
Veo 3.1	Longer production clips	Native audio, 1080p, strong physics consistency	#12 on Arena video, not the quality leader	Google AI Pro / Ultra
Kling 3.0 / 3.0 Turbo	Fast iteration at lower cost	Native 4K, 60fps, 15-second clips; Turbo shipped June 17	Outside the top 16 on Arena text-to-video	From $10/mo
Luma Ray 3	Photoreal scenes	Strong realism for landscapes	Smaller community	Free / from $9.99/mo

Runner-up and alternatives: Gemini Omni 1.1 Flash is the runner-up and the pick for text-to-video and long takes, with 40-second scene extension and 4K output, and Gemini Omni Flash is the cheaper Google option for 10-second clips. Alibaba's Wan 3.0 is third and the one to use when the source material is a document, a spreadsheet or a slide deck. Dreamina Seedance 2.5 is the ByteDance pick, and Veo 3.1 is the pick inside the Gemini app and Vertex AI when you want native audio at 1080p. OpenAI retired the Sora 2 consumer app on April 26, 2026 and removed the Sora 2 models and the Videos API from its developer platform on September 24, 2026. We track the retirement dates for older models across every provider in a separate running list.

What changed this month

The video crown moves from Gemini Omni 1.1 Flash to MiniMax H3. Since the last update Artificial Analysis has rated Omni 1.1 Flash for the first time, and on its text-to-video board, now at version 2.0, it comes eighth at 1113, behind Wan 3.0 at 1156 and MiniMax H3 at 1137. With H3 still first on Arena's image-to-video board and second on Artificial Analysis's, behind only a model built on it, H3 now places better across the four boards we track.

Omni 1.1 Flash keeps first place on Arena's text-to-video board, still a statistical tie with Gemini Omni Flash, so it stays on the podium as the text-to-video pick, and Wan 3.0 takes third place from Dreamina Seedance 2.0. Earlier this month FLUX 3 Video and Grok Imagine Video 1.5 entered Arena's text-to-video board third and fourth. In September, Gemini Omni 1.1 Flash held first place on Arena's text-to-video board and MiniMax H3 took first place on its image-to-video board.

Category pick, October 2026

Best AI for Coding

Claude Opus 5.5

#1 Arena WebDev

→

Claude Sonnet 5.5

Value pick, #1 Terminal-Bench

→

GPT-6 Astra

#2 Arena WebDev

The best AI for coding is Claude Opus 5.5, which Anthropic shipped on September 22, though it no longer leads every board, and the model that took two of them is its own cheaper sibling.

Claude Sonnet 5.5, released September 28, leads Artificial Analysis's own Terminal-Bench 4.0 run at 63.6% against Opus 5.5's 59.6% and LiveBench Coding at 91.4 against 89.3. Opus 5.5 keeps the crown on the rest of the evidence. It leads Arena's WebDev board at 1813, ahead of GPT-6 Astra at 1786 and Claude Sonnet 5.5 at 1774. It scores 71.7 on LiveBench Agentic Coding, second only to DeepSeek V4.1-Flash, where Sonnet 5.5 manages 56.3, and on Anthropic's own harnesses it beats Sonnet 5.5 on FrontierCode v1.1, 54.4% to 52.1%, and on CursorBench 4.0, 57.8% to 55.5%. Price now favours Sonnet 5.5 at every effort level.

It is $2 / $10 per 1M tokens against Opus 5.5's $4 / $20, and Artificial Analysis now puts even a max-effort Intelligence Index task at $5.46 against Opus 5.5's $5.98; at xhigh it scores 51.9 on the index for $2.01 a task and at high 46.7 for $0.88. Opus 5.5 has also taken Arena's image-to-WebDev board at 1748.5, ahead of Sonnet 5.5 at 1740.0 and GPT-6 Astra at 1731.4; Astra matches Opus 5.5's 59.6% on Terminal-Bench 4.0 at xhigh, at $10 / $50, and Claude Fable 5.1 is fifth on WebDev and fourth on image-to-WebDev.

GPT-6.1 Sol is the cheap OpenAI option: fourth on WebDev and 56.1% on Terminal-Bench 4.0 for $0.72 an index task, though LiveBench puts it well outside the top ten on Coding, at 80.7. Gemini 4 Argon leads Google's own DeepSWE v1.1 table at 77.9%, but Artificial Analysis measures 57.1% on Terminal-Bench 4.0, Arena puts it ninth on WebDev, and since October 13 it is on the paid API and in Antigravity for Google AI Ultra subscribers.

Claude Haiku 5.5, released October 7 at $0.10 / $0.50, is the cheap Claude for high-volume sub-agent work: 33% on Artificial Analysis's Terminal-Bench 4.0 run, level with GLM 5.3 Flash, and 57.2 on LiveBench Agentic Coding, just above Sonnet 5.5, though Arena puts it only 30th on WebDev. Among open weights, Xiaomi's MiMo-V2.6-Pro takes 34.8% on Terminal-Bench 4.0 under plain MIT at $0.13 an index task, GLM-5.3 is still the strongest open model on that harness at 41.9%, and Kimi K3 is the highest-placed open model on Arena WebDev, fourteenth at 1654, on only 12.6% of the harness.

Artificial Analysis has retired both its Coding Index and its Coding Agent Index, so the two composite coding rankings this page used to quote no longer exist.

Model	Best For	Strength	Weakness	Price (per 1M tokens)
Claude Opus 5.5	Best coding overall	#1 Arena WebDev (1813) and #1 Intelligence Index (57.6, v4.3.2); 71.7 LiveBench Agentic Coding, 15 points above Sonnet 5.5	Second to Sonnet 5.5 on Terminal-Bench 4.0 (59.6% to 63.6%) and LiveBench Coding (89.3 to 91.4)	$4 / $20
Claude Sonnet 5.5	Value pick for coding	#1 Terminal-Bench 4.0 (63.6%) and #1 LiveBench Coding (91.4), #3 Arena WebDev (1774); 51.9 on the index for $2.01 a task at xhigh	56.3 LiveBench Agentic Coding, 15 points behind Opus 5.5	$2 / $10
GPT-6 Astra	Second on Arena WebDev	#2 Arena WebDev (1786) and #3 image-to-WebDev (1731.4); 59.6% Terminal-Bench 4.0 at xhigh	2.5x Opus 5.5's price; behind Sonnet 5.5 on Terminal-Bench and well behind both Claude models on LiveBench Coding	$10 / $50
Claude Fable 5.1	Highest composite score before Opus 5.5	#5 Arena WebDev and #4 image-to-WebDev; Intelligence Index 53.4; AA-Briefcase 1675	52.0% Terminal-Bench 4.0; 2.5x the price of Opus 5.5	$10 / $50
Claude Opus 5	Superseded by Opus 5.5	49.0% Terminal-Bench 4.0 and #7 Arena WebDev and #7 image-to-WebDev	Costs more per token than Opus 5.5 and scores seven points lower	$5 / $25
Claude Fable 5	Hardest long-horizon agentic work	Intelligence Index 49.6; 42.4% Terminal-Bench 4.0; 1M context	The priciest per index task on this table at $8.75; #10 on Arena image-to-WebDev	$10 / $50
Qwen3.8-Max-0902	Brief holder of Arena's WebDev board	#10 Code Arena: WebDev (1674); Intelligence Index 45.4; 2.4T parameters, 1M context	QwenCloud API only with no consumer front-end; lost the WebDev lead within days	$2 / $6
Kimi K3	Highest-placed open model on Arena	#14 Arena WebDev (1654), the best open placement on that board; 2.8T/104B active	Only 12.6% on Terminal-Bench 4.0; 1.56 TB to self-host	$3 / $15
GPT-5.6 Sol	OpenAI's cheaper flagship	Intelligence Index 47.0; 39.9% Terminal-Bench 4.0	Superseded by GPT-6 Sol, which scores higher at half the price; eval-gaming flagged by METR	$4 / $20
GPT-6 Sol	Cheap GPT-6 coding in Codex	Intelligence Index 47.6 (v4.3.2) and 43.9% Terminal-Bench 4.0, both above GPT-5.6 Sol, at half its price	#8 Arena WebDev (1688)	$2 / $10
GPT-6.1 Sol	Cheapest top-tier model per task	Intelligence Index 51.8 (v4.3.2), #4 Arena WebDev (1755) and 56.1% Terminal-Bench 4.0 at $0.72 an index task, with cached input at $0.10	80.7 on LiveBench Coding, well outside the top ten; in Codex and ChatGPT Work only, not in Chat	$2 / $10
Gemini 4 Argon	Long software tasks	77.9% DeepSWE v1.1 on Google's own figures; Intelligence Index 52.6 at $1.99 an index task	57.1% Terminal-Bench 4.0 and #9 Arena WebDev (1678)	$2 / $10 intro to Nov 30, $4 / $20 standard
Muse Spark 1.3	Cheap agentic coding	Intelligence Index 48.1 at $1.25 / $4.25, and $1.60 per index task	33.3% on Terminal-Bench 4.0; the coding wins come from Meta's own chart, measured on a partners-only max variant	$1.25 / $4.25
Muse Wonder Full	Meta's strongest model, inside Muse Code	Intelligence Index 55.8 (v4.3.2), third among distinct models; #2 GDPval-AA v2.1 (1857); 57.1% Terminal-Bench 4.0	Muse Code subscription only ($50 and $200 plans), no API; Arena has not rated it	Muse Code, from $50/mo
Muse Wonder	Meta's frontier model on the API	Intelligence Index 52.4 (v4.3.2); 52.0% Terminal-Bench 4.0; 1M context	Same $4 / $20 as Opus 5.5, which scores 5.2 points higher; $4.52 an index task	$4 / $20
Grok 4.7	Cheap value coder	46.3% CursorBench 4.0 and 71.0% DeepSWE v1.1 on SpaceXAI's own figures; Intelligence Index 46.3	24.7% on Terminal-Bench 4.0; prompts of 200K tokens and up re-bill the whole request at double	$2 / $6
Gemini 3.8 Flash	Agent coding at scale	Intelligence Index 40.9; 308.4 output tokens per second	19.7% on Terminal-Bench 4.0; intro price doubles January 1, 2027	$0.75 / $3.75
GLM 5.3 Flash	Cheapest serious coder you can host	32.8% Terminal-Bench 4.0 at $0.25 an index task; MIT, 320B/18B active	GLM-5.3 reaches 41.9% on the same harness; no consumer product	Open weights (MIT)
Claude Haiku 5.5	Cheapest Claude for sub-agents	Intelligence Index 43.4 (v4.3.2) at $0.21 an index task; 33% Terminal-Bench 4.0, level with GLM 5.3 Flash; 57.2 LiveBench Agentic Coding, above Sonnet 5.5	#30 Arena WebDev (1587); Anthropic says Sonnet 5.5 and Opus 5.5 remain better for complex agentic coding; prompts above 100K tokens cost five times as much	$0.10 / $0.50
MiMo-V2.6-Pro	Highest open Intelligence Index	46.3 on v4.3.2 and 34.8% Terminal-Bench 4.0 at $0.13 an index task; MIT, 1.02T/42B active	Only #19 on Arena WebDev (1629), five places behind Kimi K3	$0.435 / $0.87

Runner-up and alternatives: Claude Sonnet 5.5 is the runner-up and the value pick, leading Terminal-Bench 4.0 and LiveBench Coding at half Opus 5.5's token price and now at a lower cost per index task, GPT-6 Astra is third and second on Arena's WebDev board, GPT-6.1 Sol is fourth on WebDev at a fifth of Astra's token price, Claude Haiku 5.5 is the cheapest Claude for high-volume sub-agent work, and Claude Opus 5 is now superseded by a model that is both cheaper and better. Among open weights, Xiaomi's MiMo-V2.6-Pro is the leader on score and on cost, GLM-5.3 is the strongest on Artificial Analysis's Terminal-Bench 4.0, Kimi K3 is the highest-placed open model on Arena's WebDev board, and GLM 5.3 Flash is the pick for teams hosting their own, at $0.25 an index task under plain MIT. Inside IDEs, Cursor with Claude is still the most popular pairing and Claude Code is the natural pick if you live in the terminal.

What changed this month

Claude Opus 5.5 keeps the coding crown. Since the last update Claude Haiku 5.5 arrived on October 7 at $0.10 / $0.50 and takes a table row as the cheapest Claude for sub-agent work, though Arena puts it only 30th on WebDev. Artificial Analysis now prices a max-effort Sonnet 5.5 index task at $5.46, below Opus 5.5's $5.98, so Opus 5.5 holds the crown on WebDev and LiveBench Agentic Coding rather than on cost. Claude Sonnet 5.5 has entered Arena's WebDev board third at 1774, behind GPT-6 Astra at 1786, which pushes GPT-6.1 Sol to fourth and Claude Fable 5.1 to fifth, and Gemini 4 Argon, which entered at the start of the month, is now ninth. Since then Opus 5.5 has also taken Arena's image-to-WebDev board from GPT-6 Astra, and Meta's Muse Wonder arrived on October 12 with two table rows: Muse Wonder Full scores 55.8 on the index but runs only in Muse Code, and Muse Wonder costs the same $4 / $20 as Opus 5.5 for 52.4.

Argon leads Google's own DeepSWE v1.1 table at 77.9%, but on Artificial Analysis's Terminal-Bench 4.0 it scores 57.1%, behind Sonnet 5.5, Opus 5.5 and GPT-6 Astra, and its public release on October 13 did not change those scores. In September the crown moved twice, from Claude Fable 5.1 to GPT-6 Astra and then to Opus 5.5, and Claude Sonnet 5.5 took Terminal-Bench 4.0 and LiveBench Coding as the value pick.

Category pick, October 2026

Best AI for Creativity

Grok 4.7

Fewest guardrails

→

Claude Fable 5

Highest-quality prose

→

Kimi K3

Fiction and voice

The best AI for unfiltered, on-trend creative work is Grok 4.7, and we want to be exact about why. This pick is about the product line, not the prose quality.

Grok carries the fewest content restrictions of any frontier model and the only native real-time X integration, which makes it the one model that will engage with edgy, topical or deliberately provocative briefs that the others decline. SpaceXAI shipped Grok 4.7 on September 21 to the Grok API, Cursor and Grok Build, and since October 3 it also powers Expert mode on grok.com, plus Fast mode for paid subscribers. It is not the best writer, though the picture is more mixed than it was.

EQ-Bench has now rated Grok 4.7 and puts it eighth at 2006.7, inside its top ten and far above Grok 4.5 in 52nd, but Arena's human voters put 4.7 only 74th on creative writing, behind Grok 4.6 in 54th and the older Grok 4.20-beta1 in 23rd. If you are picking on output quality alone, Gemini 4 Argon now leads Arena's creative-writing board, on Google AI Pro since October 13, with Claude Opus 5.5 second and Claude Fable 5 third, while EQ-Bench puts GPT-6 Astra first at 2173.3, Claude Fable 5.1 second and Kimi K3 fifth. Choose Grok for what it will let you make, not for how well it writes.

Model	Best For	Strength	Weakness	Price
Grok 4.7	Unfiltered, opinionated, on-trend	Fewest content restrictions, native real-time X grounding	#74 on Arena creative writing, though #8 on EQ-Bench (2006.7)	$2 / $6 API; $30/mo SuperGrok app
Claude Fable 5	Highest-quality creative prose	#3 Arena creative writing, #1 LiveBench Language	Cautious guardrails on edgy briefs	$10 / $50 API
Kimi K3	Fiction and distinctive voice	#5 EQ-Bench Creative Writing at 2082.3	Only #28 on Arena creative writing	$3 / $15
Claude Opus 5	Long-form structured creativity	Holds long threads and self-edits; Intelligence Index 50.8, now behind six newer models	Most cautious of the group	$20/mo Pro; $5 / $25 API
Gemini 3.1 Pro	Multimodal creative	Strong text, image and video chain	Needs Google AI Pro in the Gemini app since October 9	AI Pro / $2.00-$4.00 API in
Grok Imagine (Spicy Mode)	NSFW / adult creative	Most permissive image generation	Niche use case	$30/mo SuperGrok

Runner-up and alternatives: Claude Fable 5 is the runner-up and the right pick if quality matters more than freedom, Kimi K3 is the pick for fiction, and Claude Opus 5 is the pick for creative projects that run across many turns. For adult creative work, Grok Imagine Spicy Mode is still the only frontier-grade option.

What changed this month

Grok keeps this pick on Grok 4.7, and nothing about the product's permissiveness or its live X access changed, which is what the pick rests on. Arena's creative-writing board moved around it: Gemini 4 Argon, announced September 30, went in at first, Claude Opus 5.5 is now second and Claude Fable 5 third, and Grok 4.7 has slipped to 74th. EQ-Bench has not changed and still puts 4.7 eighth. The app the permissiveness argument is really about caught up on October 3, when Grok 4.7 started powering Expert mode on grok.com. In September, Grok 4.7 replaced Grok 4.6 as SpaceXAI's flagship on September 21 and both creative-writing boards rated it.

Category pick, October 2026

Best AI for Accuracy & Research

Claude Fable 5.1

Highest factual accuracy

→

GPT-6.1 Sol

Best value, $0.72/task

→

Claude Opus 5.5

Best when wrong costs most

The best AI for accuracy and research is Claude Fable 5.1, which holds the highest factual accuracy Artificial Analysis has measured, 67% on AA-Omniscience.

That is the exact column this category is decided on. Claude Opus 5.5 beat Fable 5.1 on almost everything else when it arrived on September 22, including Humanity's Last Exam at 61.4% to 59.1% and the hallucination-penalised AA-Omniscience index at 46.4 to 43.5, but on raw factual accuracy it reads 66% to Fable 5.1's 67%. Treat that one-point gap as a tie and read the two together: Opus 5.5 is the better bet when a wrong answer is worse than no answer, Fable 5.1 when you want the most facts right.

The value pick is now GPT-6.1 Sol, which answers 62% correctly with an AA-Omniscience index of 41.5 at $2 / $10 per 1M tokens and $0.72 an index task, and scores 98.5% on ARC Prize's ARC-AGI-1 for $0.06 a task. That beats Gemini 3.1 Pro, our value pick until this update, on every accuracy measure Artificial Analysis publishes: it reads 55% raw accuracy and an index of 31.9.

Gemini 3.1 Pro keeps its case where the answer has to be current, because it grounds answers in Google Search and is in the Gemini app (Google AI Pro and above since October 9), while GPT-6.1 Sol runs in ChatGPT Work, Codex and the API rather than in ChatGPT's chat window. Gemini 4 Argon is the model to watch: Artificial Analysis measures a 15% hallucination rate, the lowest among leading models, because it declines rather than guesses, but it answers only 50% correctly; since October 13 it is in the Gemini app on Google AI Pro and above.

On grounded search specifically, Arena's search leaderboard is led by OpenAI: GPT-5.6 Sol tops it at 1257, with Claude Fable 5 fifth and Gemini 3.1 Pro grounding ninth. On novel reasoning the crown sits elsewhere, with GPT-6 Astra leading ARC-AGI-2 and ARC-AGI-3. Claude Opus 5 holds no accuracy crown because Artificial Analysis measures its hallucination rate at 61%.

Model	Best For	Key Benchmark	Weakness	Price
Claude Fable 5.1	Highest measured factual accuracy	67% factual accuracy on AA-Omniscience, the highest Artificial Analysis has measured, one point above Claude Opus 5.5	No cheap tier; unrated on Arena's search leaderboard; Opus 5.5 leads Humanity's Last Exam and the AA-Omniscience index	$10 / $50
Claude Opus 5.5	When a wrong answer is worse than none	#1 AA-Omniscience index (46.4) and #1 Humanity's Last Exam (61.4%); 66% raw accuracy	One point behind Fable 5.1 on raw accuracy	$4 / $20
GPT-6.1 Sol	Best value	62% raw accuracy and an AA-Omniscience index of 41.5 at $0.72 an index task; 98.5% ARC-AGI-1 at $0.06 a task	54% hallucination rate; in ChatGPT Work and Codex, not in Chat	$2 / $10
Gemini 4 Argon	Fewest hallucinations	15% hallucination rate on AA-Omniscience, the lowest among leading models	Only 50% raw accuracy	$19.99/mo Google AI Pro; $2 / $10 API intro
Gemini 3.1 Pro	Grounded search; Gemini app on AI Pro from Oct 9	Native Google Search grounding; 98% ARC-AGI-1 at $0.52/task	55% raw accuracy on AA-Omniscience, below GPT-6.1 Sol; #9 on Arena search	$2.00-$4.00 / $12.00-$18.00 (tiered)
Claude Fable 5	Grounded search	#5 on Arena's search leaderboard, behind GPT-5.6 Sol	No single cheap tier	$10 / $50 (Fable 5)
GPT-5.6 Sol	Novel reasoning	#4 ARC-AGI-2 at 92.5%, behind GPT-6 Astra (95%), GPT-6.1 Sol (94.2%) and Claude Opus 5.5 (93.3%)	Scheming flagged by METR	$4 / $20
Claude Opus 5	Hardest unseen problems	30.2% ARC-AGI-3 on the standard harness, second to GPT-6 Astra (ARC Prize)	Artificial Analysis measures a 61% hallucination rate	$5 / $25
Qwen 3.7 Max	Frontier accuracy at value pricing	92.4 GPQA Diamond, 200 free requests/day	API-only, no chat front-end	$1.25 / $3.75 promo; $2.50 / $7.50 list
Claude Opus 4.6	Honesty under pressure	#1 on Scale SEAL's MASK board at 96.28; Anthropic holds the top 5	Superseded as a flagship	Legacy Anthropic model

Runner-up and alternatives: Claude Opus 5.5 is the runner-up and the pick when a wrong answer costs more than no answer, GPT-6.1 Sol is the value pick at $0.72 an index task, Gemini 3.1 Pro is the pick when answers need Google Search grounding or you want it inside the Gemini app (Google AI Pro and above since October 9), Claude Opus 4.6 sweeps the honesty-under-pressure board, and Qwen 3.7 Max is the value pick at the frontier.

What changed this month

The accuracy crown stays with Claude Fable 5.1, at 67% raw factual accuracy against Claude Opus 5.5's 66%, and the value pick moved. GPT-6.1 Sol now beats Gemini 3.1 Pro on every accuracy measure Artificial Analysis publishes, 62% raw accuracy against 55% and an AA-Omniscience index of 41.5 against 31.9, and ARC Prize scores it 98.5% on ARC-AGI-1 for $0.06 a task against Gemini 3.1 Pro's 98% for $0.52, so it takes the value card. Gemini 3.1 Pro stays in the table for grounded search and the Gemini app, where it needs Google AI Pro since October 9.

Gemini 4 Argon, announced September 30, brings the lowest hallucination rate Artificial Analysis has measured among leading models, 15%, but it answers only half its questions correctly, so its public release on October 13 changes nothing at the top. Two older lines are corrected as well: Claude Opus 5's hallucination rate now reads 61% rather than 50%, and it no longer leads ARC-AGI-3. In September, Opus 5.5 took the hallucination-penalised index and Humanity's Last Exam, which is why this block now splits the two Claude models rather than crowning one outright.

Category pick, October 2026

Best AI for Problem Solving

GPT-6 Astra

Reasoning leader

→

Claude Opus 5

Agentic chains

→

GPT-6.1 Sol

Value pick, $2 / $10

The best AI for hard problem solving is GPT-6 Astra, which took the reasoning boards off GPT-5.6 Sol within days of launching.

It leads LiveBench Reasoning at 92.65 and ARC-AGI-2 at 95%, the highest anyone has scored against that board's 100% human panel, and sits fourth on LiveBench Mathematics at 96.81, behind Claude Opus 5.5 at 97.08, Claude Fable 5.1 at 97.01 and GPT-6.1 Sol at 96.83. It also reports 97.6% on FrontierMath Tier 4 v2, a number worth carrying with Epoch AI's disclosure that OpenAI funded the benchmark and holds exclusive access to part of it. The catch is price and reach: $10 / $50 per 1M tokens against GPT-6.1 Sol's $2 / $10, and it needs a paid ChatGPT plan.

GPT-6.1 Sol, released September 29, is the value alternative and very nearly a co-leader: 92.63 on LiveBench Reasoning, 0.02 behind Astra, and 94.2% on ARC-AGI-2, 0.8 points behind, for $0.25 a task against Astra's $1.12. Astra keeps the crown because it still leads both boards and ARC-AGI-3, where it scores 62.7% on the standard harness against GPT-6.1 Sol's 52.7%. Gemini 4 Argon leads Arena's math category, but on only a few hundred votes, and neither LiveBench nor ARC Prize has rated it. GPT-5.6 Sol, the old value pick, is now fifth on LiveBench Reasoning and fourth on ARC-AGI-2 at 92.5%.

Qwen 3.7 Max is the budget pick for competition-style problems at 97.1 on the February 2026 HMMT index and 44.5 on Apex, with 200 free model requests a day. Claude Opus 5 remains the alternative for long agentic reasoning chains, though Astra has taken ARC-AGI-3 from it as well.

Model	Best For	Key Benchmark	Weakness	Price
GPT-6 Astra	Hardest math, science and reasoning	#1 LiveBench Reasoning (92.65), #1 ARC-AGI-2 (95%), #1 ARC-AGI-3	Needs a paid ChatGPT plan; 5x GPT-6.1 Sol's price	$10 / $50
Claude Opus 5	Long agentic reasoning chains	1662 on AA-Briefcase v1.1 and 1724 on GDPval-AA v2.1, fourth among distinct models on both; 30.2% ARC-AGI-3	Astra now leads ARC-AGI-3; 61% hallucination rate; now a legacy model	$5 / $25
GPT-5.6 Sol	Value STEM flagship	#4 ARC-AGI-2 (92.5%); #5 LiveBench Reasoning (91.65), Mathematics 96.20	FrontierMath still unpublished; scheming flagged by METR	$100/mo ChatGPT Pro; API $4 / $20
GPT-6.1 Sol	Value pick, near-Astra reasoning	#2 LiveBench Reasoning (92.63), #2 ARC-AGI-2 (94.2%), #3 LiveBench Mathematics (96.83); Intelligence Index 51.8 at $0.72 an index task	52.7% on ARC-AGI-3 against Astra's 62.7%; in ChatGPT Work and Codex, not in Chat	$2 / $10
Qwen 3.7 Max	Competition math on a budget	97.1 HMMT 2026 Feb, 44.5 Apex, 200 free requests/day	API-only	$1.25 / $3.75 promo; $2.50 / $7.50 list
Claude Fable 5	Math inside a coding workflow	#3 on Arena's math subcategory (1522), behind Gemini 4 Argon and Claude Opus 5; 96.0 LiveBench Mathematics	Priciest option here	$10 / $50
GLM 5.3 Flash	Open-weight problem solving	Intelligence Index 41.8 under MIT, almost eight points above GLM-5.2 at less than half the size; 320B/18B active, 1M context	Needs a multi-GPU server, not a workstation; GLM 5.3 scores higher on Z.ai's own figures and is downloadable since August 28, but at more than twice the size and under a bespoke licence	Open weights (MIT)

Runner-up and alternatives: GPT-6.1 Sol is the runner-up and the value pick at $2 / $10, Claude Opus 5 is the natural pick for long-chain agentic reasoning, Qwen 3.7 Max is the budget pick, and GLM 5.3 Flash is the open-weight pick. Our dedicated guide to the best AI for math covers the task-by-task split and the free options.

What changed this month

GPT-6 Astra keeps the problem-solving crown, by less than a point. ARC Prize has now scored GPT-6.1 Sol, and it takes 94.2% on ARC-AGI-2, second to Astra's 95% and ahead of Claude Opus 5.5's 93.3%, for under a quarter of Astra's cost per task. Together with its 92.63 on LiveBench Reasoning, 0.02 behind Astra, that makes GPT-6.1 Sol a near co-leader, but Astra still leads both boards and ARC-AGI-3, so the crown does not move. GPT-5.6 Sol drops to fourth on ARC-AGI-2. Gemini 4 Argon, announced September 30, leads Arena's math category on a few hundred votes but has no LiveBench or ARC Prize score yet, and Claude Fable 5, which led that Arena category, is now third. In September, Astra took this crown from GPT-5.6 Sol after its September 3 launch, and LiveBench Mathematics passed to Claude Opus 5.5.

Category pick, October 2026

Best AI Agent

Gemini Spark

Cloud-resident

→

Claude Cowork

Desktop-resident

→

OpenAI dots

Always-on, runs on GPT-6 Astra

The best AI agent is Gemini Spark if you want the agent in the cloud and Claude Cowork if you want it on your desktop.

These are joint picks rather than a first and a second: Spark runs in a Google Cloud VM, so it keeps working while your laptop is shut, wired into Gmail, Docs and, since early September, Google Photos; Cowork runs on your Mac or Windows machine and drives your local apps and your screen. Spark is no longer the only agent that keeps working in the cloud.

SpaceXAI's Grok Bot (August 11) gives each account a persistent cloud computer that signs into your apps, from a $20/month Cursor Pro plan; Meta's Muse (September 8) runs errands from a cloud VM of its own and is free up to 100 million tokens a week; and OpenAI dots (September 29) run on GPT-6 Astra with their own cloud computer and browser, from the $100/month Pro plan, which does not include them in the EEA, UK or Switzerland. No independent board rates these products against each other, so the choice is about where the work has to happen rather than which one scores higher.

Price is no longer a tiebreaker either: Spark now reaches the $19.99/month Google AI Pro tier in the US and more than 160 other countries, and Cowork is included at no extra charge on every paid Claude plan from $20/month. ChatGPT Codex Mobile (May 14) is the alternative for coding-agent work. Read the full Gemini Spark vs Claude Cowork comparison.

Agent	Best For	Where It Runs	Strength	Price
Gemini Spark	24/7 cloud tasks, Workspace workflows	Google Cloud VM (always-on)	Always-on, deep Workspace integration	From $19.99/mo Google AI Pro
Claude Cowork	Desktop, app-driving, design + code	Your Mac/Windows desktop	Drives local apps, sees your screen	$20/mo Claude Pro
ChatGPT Codex Mobile	Coding agent on phone	OpenAI cloud + iOS/Android	Approve diffs and redirect work from phone	Included in ChatGPT plans
Grok Bot	Browser jobs across logged-in apps	One persistent SpaceXAI cloud computer shared by all your Bots	Signs into apps with your own logins and keeps going with your laptop closed	From $20/mo Cursor Pro, or a linked SuperGrok plan
OpenAI dots	Ongoing responsibilities, not one-off tasks	OpenAI cloud computer and browser	Runs on GPT-6 Astra; keeps working with your computer off; reachable in ChatGPT, Slack and by voice	From $100/mo ChatGPT Pro (not EEA/UK/CH); Business Premium
Meta Muse	Free personal errands	A Muse Secure VM in Meta's cloud	Free up to 100M tokens a week; a separate Sentinel agent approves what it sends out	Free; $20/mo Power, $100/mo Maximum

Runner-up and alternatives: Gemini Spark and Claude Cowork are joint picks split by where the agent runs, not a first and a second. OpenAI dots are the pick if you already pay for ChatGPT Pro and want an agent on OpenAI's flagship, Grok Bot is the pick for browser jobs across your logged-in apps, from a $20/month Cursor Pro plan, Meta's Muse is the one to try for free, and ChatGPT Codex Mobile is the pick for coding agents. Google AI Ultra at $99.99 or $199.99 a month now buys Spark higher usage limits rather than access, which starts on the $19.99 Pro tier.

What changed this month

Gemini Spark (cloud) and Claude Cowork (desktop) remain joint picks, because this category crowns products and no new agent product shipped since the last update. The model layer moved. Claude Sonnet 5.5 has entered Arena's Agent Arena third, behind Claude Fable 5.1, which still tops it with 17.6% confirmed success, and Claude Opus 5.5, and it has taken the lead on Artificial Analysis's AA-Briefcase v1.1 at 1823, with Opus 5.5 second at 1807. Opus 5.5 still leads GDPval-AA v2.1 at 1866, with Meta's new Muse Wonder Full second at 1857 and Sonnet 5.5 third at 1838. Gemini 4 Argon, announced September 30, leads Artificial Analysis's AutomationBench at 77.5% and has a 15.4% confirmed-success rate on the Agent Arena, above Opus 5.5's 14.1%, but sits tenth there overall and since October 13 it is on the paid API and in the Gemini app on Google AI Pro.

Among open models on the Agent Arena, Kimi K3 is now the highest at fourteenth, ahead of DeepSeek V4.1-Flash at seventeenth and Tencent's Hy4 Preview at nineteenth. In September, Grok Bot, Meta's Muse and OpenAI dots joined Gemini Spark as always-on cloud agents, and Spark reached the $19.99/month Google AI Pro tier.

Fello AI running on Mac, iPhone and iPad

AI one shortcut away

Press ⌥ Space on your Mac, ask, and keep working. Also on iPhone and iPad.

Free to start, 4.7★ across 27,000+ reviews.

Download Fello AI

Use case guide, October 2026

Best AI for Students

GPT-6 Luna

Essays & research

→

Gemini 3.6 Flash

STEM + PDFs

→

Claude Sonnet 5.5

Writing & editing

The best AI for students is GPT-6 Luna inside ChatGPT for general coursework and Gemini 3.6 Flash for STEM and multimodal study, with Qwen 3.7 Max as the API alternative for harder problem sets (200 free requests a day) and Claude Sonnet 5.5 as the free alternative for essay editing.

Most students don't need to pay. Since October 8, ChatGPT Free and Go have been moving from GPT-5.6 Luna to GPT-6 Luna, which Artificial Analysis scores 38.1 on its index against 37.3; the free plan has unlimited text chats, while file uploads and image tools stay capped. Claude Sonnet 5.5 is free on claude.ai, where free accounts can also pick Claude Haiku 5.5 (October 7), the Claude model that uses the least of your usage limit.

Gemini changed on October 9: the free Gemini app now runs on Auto, which sends most prompts to Flash-Lite, so choosing Gemini 3.6 Flash yourself needs Google AI Plus, though it stays free in Google AI Studio. For step-by-step working on the hardest math, GPT-6 Astra reports 97.6% on FrontierMath Tier 4 v2 but is not on ChatGPT's free tier; Qwen 3.7 Max is the value alternative at 97.1 HMMT 2026 February with API pricing at $1.25 / $3.75 on its current 50% promo ($2.50 / $7.50 list).

Task	Best Model	Why	Free?	Alternative
Essays & coursework	GPT-6 Luna	ChatGPT's Free and Go model, rolling out since October 8; unlimited text chats	Yes	Claude Sonnet 5.5 (free Claude)
STEM problem-solving	GPT-6 Sol / Qwen 3.7 Max	ChatGPT's paid default; Astra reports 97.6% FrontierMath Tier 4 v2 / 97.1 HMMT 2026 Feb	Plus paid / Qwen API paid	Gemini 3.6 Flash (AI Plus since Oct 9)
Research & accuracy	Gemini 3.1 Pro	98% ARC-AGI-1 at $0.52/task, native Google Search grounding	AI Pro (Gemini app) since Oct 9	Claude Opus 5
Writing editing	Claude Sonnet 5.5	Free on claude.ai since September 28 and second only to Opus 5.5 on the Intelligence Index; Claude Fable 5.1 is the quality leader	Yes (Claude free)	Claude Fable 5.1
Multimodal study (PDFs, slides, images)	Gemini 3.6 Flash	1M context via AI Studio; Gemini app needs AI Plus since Oct 9	Free in AI Studio	NotebookLM (Google)

Runner-up and alternatives: Claude Sonnet 5.5 (free) is the runner-up for essay writing and editing. Gemini 3.6 Flash (free in Google AI Studio, Google AI Plus in the Gemini app) is the runner-up for multimodal study and PDF ingestion. DeepSeek V4 is the runner-up for problem-solving on a strict zero-cost budget.

What changed this month

Two changes for students. ChatGPT's free model is moving from GPT-5.6 Luna to GPT-6 Luna, rolling out to Free and Go since October 8, so GPT-6 Luna takes the essays row. And since October 9 the free Gemini app runs on Auto, which sends most prompts to Flash-Lite, so picking Gemini 3.6 Flash in the app needs Google AI Plus.

Anthropic also released Claude Haiku 5.5 on October 7, which free Claude accounts can pick, and Claude Sonnet 5.5 is still free on claude.ai. Gemini 4 Argon, public since October 13, comes with Google AI Pro, so US students on Google's free year of AI Pro get it, while the free plan and the Google AI Plus student offer elsewhere do not. In September, Claude Sonnet 5.5 replaced Sonnet 5 as the free writing and editing pick.

Use case guide, October 2026

Best AI for Work & Professionals

GPT-6

Daily knowledge work

→

Claude Opus 5.5

Coding & writing

→

Gemini 3.1 Pro

Research & briefings

The best AI for professional work is GPT-6 (ChatGPT's default since October 7, Sol on paid plans) for daily knowledge work, Claude Opus 5.5 for coding and high-stakes writing, and Gemini Spark for 24/7 agentic workflows.

Most professionals get the most out of running two paid subscriptions (ChatGPT Plus at $20/month plus Claude Pro at $20/month, total $40/month). Or keep one bill: Fello AI covers ChatGPT, Claude and Gemini for everyday work. For agentic work that runs while you sleep, Gemini Spark is the always-on cloud agent that lives inside Google Workspace, and since July 30 it reaches the $19.99/month Google AI Pro tier as well as Google AI Ultra; OpenAI dots, Grok Bot and Meta's Muse are the alternatives, covered in the agents section above.

Use Case	Best Model	Key Stat	Price	Alternative
Daily knowledge work	GPT-6	ChatGPT's default since October 7; GPT-6 Sol on Plus	$20/mo ChatGPT Plus	Claude Opus 5.5
Legal, finance & tax research	Gemini 4 Argon	#1 Vals Index (68.9%) and #1 AutomationBench-AA (77.5%)	$19.99/mo Google AI Pro	Claude Opus 5.5
ChatGPT Work & Codex	GPT-6.1 Sol	In ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu since September 29, at $2 / $10 in the API	$20/mo ChatGPT Plus	GPT-6 Luna for volume work
Coding (proprietary)	Claude Opus 5.5	#1 Arena WebDev (1813) and #1 Intelligence Index (57.6), at $4 / $20	$20/mo Claude Pro	Claude Sonnet 5.5, which leads Terminal-Bench 4.0 at half the token price
Coding (cost-effective)	Claude Sonnet 5.5	#1 Terminal-Bench 4.0 (63.6%) and #1 LiveBench Coding (91.4); $0.88 an index task at high effort	$2 / $10	Qwen3.8-Max-0902; DeepSeek V4.1-Flash
Research & briefings	Gemini 3.1 Pro	98% ARC-AGI-1 at $0.52/task, Google grounding	Google AI Pro / Ultra	Claude Opus 5.5
Hard math, physics, finance modelling	GPT-6 Astra	#1 LiveBench Reasoning and ARC-AGI-2 (95%); reports 97.6% FrontierMath Tier 4 v2	$100/mo ChatGPT Pro	GPT-6.1 Sol; Qwen 3.7 Max
Always-on agent workflows	Gemini Spark	Always-on cloud agent inside Google Workspace	From $19.99/mo Google AI Pro	Claude Cowork
Live news, X-context creative	Grok 4.7	Intelligence Index 46.3 + native X grounding; in the Grok app since October 3	$30/mo SuperGrok	Gemini 3.1 Pro
All-in-one consolidation	Fello AI	ChatGPT + Claude + Gemini + Grok + DeepSeek	From $11.99/mo	Pay each vendor separately

Runner-up and alternatives: for most professional teams, Claude Opus 5.5 is the runner-up to GPT-6 for daily work and the coding leader at $4 / $20, with Claude Sonnet 5.5 the cheaper coding pick that leads Terminal-Bench 4.0 and GPT-6 Astra second on Arena's WebDev board. Gemini 3.1 Pro is the runner-up for research-heavy roles, and Gemini Spark is the pick if you can put a cloud agent to work on long tasks inside Google Workspace, with OpenAI dots the alternative for ChatGPT Pro subscribers.

What changed this month

The daily knowledge work row moves from GPT-5.6 to GPT-6, which became ChatGPT's default on October 7. Gemini 4 Argon takes a new legal, finance and tax row: it leads the Vals Index of that work and Artificial Analysis's AutomationBench, and since October 13 it is in the Gemini app on Google AI Pro at $19.99/month. In September, Claude Opus 5.5 took the proprietary coding row at a lower price than Opus 5, Claude Sonnet 5.5 took the cost-effective coding row, and GPT-6.1 Sol took the ChatGPT Work and Codex row on September 29.

Pricing Comparison

AI Model Pricing in October 2026

From $0 free tiers to $199.99/month Google AI Ultra. The price that matters most on this table is Claude Opus 5.5 at $4 / $20. Since September 22 the highest-scoring model any independent evaluator has measured costs less than the Claude Opus 5 it replaced ($5 / $25), and cache reads fall 60% to $0.20. Claude Fable 5.1 and GPT-6 Astra both list at $10 / $50 and both score below it.

The busiest rate is now $2 / $10, shared by Claude Sonnet 5.5, GPT-6 Sol and GPT-6.1 Sol. GPT-6.1 Sol is the cheapest route to a top-six Intelligence Index score, at $0.72 an index task. With Sonnet 5.5 the effort setting decides the bill: an index task costs $0.88 at high effort and $5.46 at max, now just under Opus 5.5's $5.98, and Anthropic halved its cache reads to $0.10 on October 7. Google's Gemini 4 Argon also starts at an introductory $2 / $10 ($4 / $20 standard), until November 30, and since October 13 anyone on the paid API can buy it.

Token price and task cost can drift apart. Grok 4.7 lists at $2 / $6, the same as Grok 4.6, yet an index task costs $2.73 against $1.86, because 4.7 writes roughly twice the output. Mistral Large 4 (public preview since October 6) lists at $1.36 / $4.18, though Mistral's own model page shows half that, and it is verbose enough that an index task costs $1.13 for a score of 38.4. At the frontier the value pick is Meta's Muse Spark 1.3 at $1.25 / $4.25, or $1.60 an index task for 48.1.

At the cheap end, GPT-6 Luna is the cheapest model per task on this page: $0.10 / $0.50, or $0.07 an index task. Claude Haiku 5.5 (October 7) matches that token price on prompts up to 100,000 tokens and scores 43.4 to Luna's 38.1, but it writes so much more that an index task costs $0.21 at max effort, or $0.12 for 41.3 at xhigh; above 100,000 tokens it rises to $0.50 / $2.50. Among open models sold by the token, GLM 5.3 Flash stays the pick at $0.15 / $0.50 ($0.25 an index task for 41.8), ahead of DeepSeek V4.1-Flash at $0.15 / $0.60 off-peak and $0.30 / $1.20 at peak ($0.27 for 39.5). Xiaomi's MiMo-V2.6-Pro is cheaper still at $0.13 an index task for 46.3, but only if you host its 1.02 trillion parameters yourself. For a deeper breakdown see our full AI Pricing Comparison Guide.

Model	Input (per 1M)	Output (per 1M)	Context	Free Access?
GPT-6 Astra	$10.00	$50.00	1,050,000 (922,000 max input)	In Chat on Pro, Business and Enterprise since September 4; Plus gets it in ChatGPT Work and Codex rather than Chat; no free tier; live in the API
GPT-6 Sol	$2.00	$10.00	1,050,000 (922,000 max input)	ChatGPT Chat on Plus, Pro, Business and Enterprise since October 7; Work & Codex; API
GPT-6.1 Sol	$2.00	$10.00	1,050,000 (922,000 max input)	ChatGPT Work & Codex on Plus, Pro, Business, Enterprise and Edu since September 29; API; not in Chat; cache reads $0.10
GPT-6 Luna	$0.10	$0.50	1,050,000 (922,000 max input)	ChatGPT Chat on Free and Go, rolling out since October 8; Work & Codex on paid plans; API
GPT-5.5	$5.00	$30.00	1M (400K in Codex)	ChatGPT Free; API paid
GPT-5.5 Pro	$30.00	$180.00	1M	ChatGPT Pro from $100/mo ($200 and $500 higher-usage tiers)
GPT-5.6 Sol	$4.00	$20.00	Not published	Live in ChatGPT, Codex & API (July 9); promotional rate through at least November 21, 2026
GPT-5.6 Terra	$2.00	$12.00	Not published	Live in ChatGPT, Codex & API (July 9)
GPT-5.6 Luna	$0.20	$1.20	Not published	Live in ChatGPT, Codex & API (July 9)
Claude Opus 5.5	$4.00	$20.00	1M	Claude Pro/Max/Team; API paid; cache reads $0.20
Claude Opus 5	$5.00	$25.00	1M	Legacy model at Anthropic; Pro/Max/API
Claude Opus 4.8	$5.00	$25.00	1M	Legacy model at Anthropic; Pro/Max/API
Claude Fable 5.1	$10.00	$50.00	1M	Cache reads $0.25; in Max/Team Premium (~50% of usage limits); Pro/Team Standard via credits. Mythos 5.1 bills identically, by invitation only
Claude Fable 5	$10.00	$50.00	1M	Permanent in Max/Team Premium (~50% of usage limits); Pro/Team Standard via credits
Claude Sonnet 5.5	$2.00	$10.00	1M	Free on claude.ai; Pro/Max/Team; API paid; cache reads $0.10 since October 7
Claude Sonnet 5	$2.00	$10.00	1M	Legacy model since September 28; the fallback for flagged cyber requests; API paid
Claude Sonnet 4.6	$3.00	$15.00	1M	API paid (superseded by Sonnet 5)
Claude Haiku 5.5	$0.10 (≤100K) / $0.50 (>100K)	$0.50 (≤100K) / $2.50 (>100K)	1M	Free, Pro, Max, Team and Enterprise on Claude.ai (web, iOS, Android); API paid; cache reads $0.01
Gemini 3.1 Pro	$2.00 (≤200K) / $4.00 (>200K)	$12.00 (≤200K) / $18.00 (>200K)	1M	Gemini app on Google AI Pro and above since Oct 9; API paid
Gemini 4 Argon	$2.00 intro to Nov 30 / $4.00 standard ($0.10 cached, intro)	$10.00 intro / $20.00 standard	1M, shared by input and output	Paid API since October 13; Gemini app on Google AI Pro and Ultra
Gemini 3.8 Flash	$0.75 intro / $1.50 from Jan 1, 2027	$3.75 intro / $7.50 from Jan 1, 2027	1M	AI Studio, Antigravity, Gemini Enterprise + paid API; in the Gemini app reported for AI Pro/Ultra
Gemini 3.7 Flash	$0.75 intro / $1.50 from Jan 1, 2027	$3.75 intro / $7.50 from Jan 1, 2027	1M	AI Studio, Android Studio, Antigravity + paid API; in the Gemini app via Spark only (AI Pro/Ultra, excludes EEA/UK/CH/Nigeria)
Gemini 3.6 Flash	$0.75 intro / $1.50 from Jan 1, 2027	$3.75 intro / $7.50 from Jan 1, 2027	1M	Gemini app on AI Plus and above since Oct 9; AI Studio; free API tier + paid API
Gemini 3.5 Flash-Lite	$0.30	$2.50	1M	AI Studio; free API tier + paid API
Qwen3.8-Max-0902	$2.00 ($0.17 explicit cache hit, $0.25 implicit)	$6.00	1M (128K output)	QwenCloud API only; no consumer chat front-end; open weights only as the text-only Qwen3.8-2.4T-A95B it is built on
Qwen 3.7 Max	$1.25 promo / $2.50 list	$3.75 promo / $7.50 list	1M	200 free requests/day; API paid beyond that
MiniMax M3	$0.30 (50% off $0.60)	$1.20 (≤512K)	1M	Open weights; hosting costs apply
LongCat-2.0	Provider-dependent	Provider-dependent	1M	Open weights (MIT); hosting costs apply
NVIDIA Nemotron 3 Ultra	Provider-dependent	Provider-dependent	1M	Open weights (OpenMDW); hosting costs apply
Qwen 3.5 (open-weight)	Self-host / Together	Self-host / Together	1M	Open weights; hosting costs apply
Nex-N2-Pro	Self-host / providers	Self-host / providers	1M	Open weights (Apache 2.0); hosting costs apply
Rio 3.5 Open 397B	Self-host / providers	Self-host / providers	1M	Open weights (MIT); hosting costs apply
Grok 4.3	$1.25	$2.50	1M	Free consumer plan; API paid
Grok 4.6	$2.00 (<200K) / $4.00 (≥200K)	$6.00 (<200K) / $12.00 (≥200K)	500K	SpaceXAI API, Cursor, Grok Build, OpenRouter, Vercel, Cloudflare
Grok 4.7	$2.00 (<200K) / $4.00 (≥200K)	$6.00 (<200K) / $12.00 (≥200K)	500K	SpaceXAI API, Cursor, Grok Build, harnesses and cloud routers; not in the Grok app
Muse Spark 1.3	$1.25 ($0.10 Contributor tier)	$4.25 ($0.20 Contributor tier)	1M	Paid API; Muse Code, Meta Model API and OpenRouter; Contributor tier trades training rights for a ~92% discount
Muse Wonder	$4.00	$20.00	1M	Paid API; Meta Model API, Muse Code and OpenRouter; Muse Wonder Full is only in Muse Code, on the $50 and $200 plans
Kimi K3	$3.00 ($0.30 cache-hit)	$15.00	1M	Free basic tier in the Kimi app; open weights on Hugging Face (Kimi K3 License)
Mistral Large 4	$1.36 ($0.68 on Mistral's model page)	$4.18 ($2.09)	512K (1M per Mistral)	Public preview API since October 6; open weights promised by the end of October, no licence named yet
Gemini Omni Flash (video)	$1.50	$17.50 (video output)	10-second clips	Gemini app / Flow; AI Studio + API
DeepSeek V4-Pro	$0.66 off-peak / $1.32 peak ($0.022 cache-hit off-peak)	$1.98 off-peak / $3.96 peak	1M	DeepSeek Chat free; API paid, peak/off-peak tiers since August 16
DeepSeek V4.1-Flash	$0.15 off-peak / $0.30 peak ($0.003 cache-hit off-peak)	$0.60 off-peak / $1.20 peak	1M	DeepSeek Chat free; API paid, peak/off-peak tiers; replaced V4-Flash on September 10
Kimi K2.7 Code	Provider-dependent	Provider-dependent	256K	Open weights; hosting costs apply
GLM-5.2	Provider-dependent	Provider-dependent	1M	Open weights; hosting costs apply
GLM 5.3	$1.40 ($0.26 cache-hit)	$4.40	1M	Z.ai API, GLM Coding Plan and ZCode; weights on Hugging Face since August 28 under the bespoke GLM 5.3 License
GLM 5.3 Flash	$0.15 ($0.03 cache-hit)	$0.50	1M	Z.ai API, GLM Coding Plan, OpenRouter; MIT weights on Hugging Face; the launch promotion ended September 9
Tencent Hy4 Preview	$0.834 ($0.042 cache-hit)	$2.501	1M+	Open weights (Apache 2.0); Tencent Cloud TokenHub, OpenRouter, WorkBuddy, CodeBuddy
Qwen3.8-Flash-Next	Provider-dependent	Provider-dependent	262K (to 1M)	Open weights (Qwen Community License 1.0); hosting costs apply
MiMo-V2.6-Pro	$0.435	$0.87	1M	Open weights (MIT); hosting costs apply
ERNIE 5.1	China-region pricing	China-region pricing	256K	Baidu free tier
Gemini Spark (agent)	Not API-priced	Not API-priced	1M (Gemini base)	Google AI Pro $19.99, AI Ultra $99.99 or $199.99/mo
Fello AI (aggregator)	Routed via app	Routed via app	Model-dependent	From $11.99/mo, free tier available

The GPT-5.5 and GPT-5.5 Pro rates above are short-context prices; OpenAI no longer publishes the specific long-context figures. The GPT-5.6 tiers are billed at 2x input and 1.5x output once a prompt passes 272K input tokens, which puts long-context Terra at $4 / $18 and Luna at $0.40 / $1.80. The GPT-6 tiers carry the same surcharge past 272K input tokens, applied to the whole request, and their cached input reads are 90% off: $1.00 for Astra, $0.20 for GPT-6 Sol and $0.01 for Luna, with GPT-6.1 Sol at 95% off, $0.10.

Grok 4.6 works the same way but bites harder: at 200,000 prompt tokens and above, SpaceXAI re-bills the entire request at the higher rate rather than only the overage, so crossing the line by a thousand tokens doubles the cost of the whole call. DeepSeek is the other rate card that needs reading twice: since August 16 both V4 models bill at peak rates from 01:00 to 04:00 and from 06:00 to 10:00 UTC on weekdays, and at exactly half that in every other hour, so the same job can cost twice as much depending on when you run it.

The cheapest way through this table is often one subscription instead of three. Fello AI puts GPT, Claude, Gemini, Grok, Perplexity and more in a single app for Mac, iPhone and iPad, from $11.99 a month in most countries.

Open-Weight Models

Best Open-Weight Models in October 2026

The best open-weight model in October 2026 is Xiaomi's MiMo-V2.6-Pro, published on September 21 under plain MIT: 1.02 trillion parameters with 42 billion active and a 1M-token context. Artificial Analysis scores it 46.3 on Intelligence Index v4.3.2, the highest open score it has ever published, ahead of GLM-5.3 at 44.8 and Kimi K3 at 43.6, at just $0.13 an index task. The catch: Arena's human voters rank it only 19th on WebDev and 26th on text.

Two other open models still lead somewhere. Kimi K3 is the highest-placed open model on Arena, 14th on WebDev and 14th on the Agent Arena. GLM-5.3 is the strongest open model on Terminal-Bench 4.0 at 41.9%, though on a bespoke Z.ai licence rather than MIT.

For teams running their own weights, the practical pick is still GLM 5.3 Flash (Z.ai, MIT): 41.8 on v4.3.2 from 320B total and 18B active. It is a separate model on a newly trained base, not a trimmed GLM 5.3, which is why it ships under MIT. A 1.02-trillion-parameter model is not a workstation proposition, and the K3 download alone is about 1.56 TB. For the cheapest tier, DeepSeek-V4.1-Flash replaced V4-Flash on September 10: MIT, natively multimodal, 39.5 on v4.3.2 and $0.15 / $0.60 off-peak.

Three more are worth watching. Mistral Large 4 (public preview since October 6) is the highest-scoring model from the US or Europe that promises open weights, at 38.4, but you cannot download it yet: Mistral says the weights follow by the end of October and has not named a licence. Alibaba's Qwen3.8-Flash-Next, 125B with only 6B active, previews the Qwen4 architecture and scores 39.8. Tencent's Hy4 Preview, 770B total and 49B active under Apache 2.0, has no Artificial Analysis rating but sits 18th on Arena WebDev.

Model	Best For	Key Benchmark	Context / License	Where To Run
MiMo-V2.6-Pro	Highest open Intelligence Index ever measured	II 46.3 (v4.3.2), above GLM-5.3 and Kimi K3, at $0.13 an index task; 34.8% Terminal-Bench 4.0; GDPval-AA v2.1 1685; 1.02T/42B active	1M / MIT	Hugging Face (XiaomiMiMo), providers, self-host
Kimi K3	Highest-placed open model on Arena	II 43.6 (v4.3.2); #14 Arena WebDev, the best open placement there; #14 Agent Arena; 2.8T/104B active	1M / Kimi K3 License	Hugging Face (96 shards, 1.56 TB), Moonshot API, providers
GLM 5.3 Flash	Best open model most teams can actually run	II 41.8 (v4.3.2), on the intelligence-versus-cost Pareto frontier at about $0.25 an index task; 320B/18B active, natively multimodal	1M / MIT	Hugging Face, Z.ai API ($0.15/$0.50), OpenRouter
GLM-5.2	Fallback with an independent track record	II 33.7 (v4.3.2); #27 Arena WebDev (1,603.2), #20 Agent Arena; 744B/40B active	1M / MIT	Z.ai, Hugging Face, OpenRouter
GLM 5.3	Open since August 28, but on a bespoke licence	II 44.8 (v4.3.2), the second-highest open score we found; same 744B base as GLM-5.2, post-trained only, published checkpoint counted at 753B; CyberGym 84.5% and Terminal-Bench 3.0 28.3 (vendor figures)	1M / GLM 5.3 License (MaaS above $10B revenue needs a Z.ai security review)	Hugging Face, Z.ai API ($1.40/$4.40), GLM Coding Plan, ZCode
DeepSeek V4.1-Flash	Best open value if you host it yourself	II 39.5 (v4.3.2) against 34.3 for the V4-Flash 0731 build it replaces; 552B, 8B active at prefill and 16B at decode	1M / MIT	DeepSeek API ($0.15/$0.60 off-peak, $0.30/$1.20 peak since September 10), local
Tencent Hy4 Preview	Largest permissively licensed model yet	770B/49B active; 2.99/4.00 on Tencent's own 203-task expert panel vs Kimi K3 2.94 and GLM 5.3 2.92 (vendor); no Artificial Analysis rating; #18 Arena WebDev (1632), #19 Agent Arena	1M+ / Apache 2.0	Hugging Face (BF16 and FP8), Tencent Cloud TokenHub, OpenRouter
Qwen3.8-Flash-Next	Preview of the Qwen4 architecture	125B total plus a 51B N-gram embedding, 6B active; 91.7 GPQA Diamond and 62.5 SWE-Bench Pro (vendor); II 39.8 (v4.3.2); #17 Arena WebDev (1633)	262K to 1M / Qwen Community License 1.0	Hugging Face, providers, self-host
LongCat-2.0	Frontier open coder trained on Chinese chips	II 33 (v4.1); 59.5% SWE-Bench Pro (vendor), 1.6T/~48B active	1M / MIT	Hugging Face, GitHub, OpenRouter
MiniMax M3	Cheap long-context multimodal	II 29.2 (v4.3.2; 44 on v4.1), 59% SWE-Bench Pro, multimodal	1M / MiniMax Community License	Hugging Face, API $0.30/1M (50% off)
Nex-N2-Pro	Strongest open coding score	II 41 (v4.1); 80.8 SWE-Bench Verified, 397B/17B active	Qwen-based / Apache 2.0	Hugging Face, providers, self-host
Kimi K2.7 Code	Strongest commercially-licensed open coder	+21.8% on Kimi Code Bench v2 vs K2.6 (vendor); 1T/32B active	256K / Modified MIT	Hugging Face, DeepInfra, providers
DeepSeek V4-Pro	Agentic real-world work	II 44 (v4.1), 1.6T/49B active	1M / MIT	DeepSeek API ($0.66/$1.98 off-peak, $1.32/$3.96 peak since August 16), local
Hy3	Newest permissive-licence entrant	II 41 (v4.1); #56 on Arena WebDev (1,508.4)	Apache 2.0	Hugging Face, providers, self-host
Inkling	Thinking Machines' first open model	II 41 (v4.1), agentic 32.3, released July 15	Open weights	Hugging Face, providers, self-host
Inkling Small	Same family at a quarter the size	II 40 (v4.1), agentic 30.8; 276B/12B active, text, image and audio in	Apache 2.0	Hugging Face (BF16 and NVFP4), providers, self-host
NVIDIA Nemotron 3 Ultra	NVIDIA-tuned, fully permissive license	II 38 (v4.1), 65-70.4 SWE-Bench Verified, 550B/55B active	1M / OpenMDW	OpenRouter, Hugging Face, AWS (8x B200 self-host)
Qwen 3.5 (397B / 17B active)	Multimodal, fast decode	88.4 GPQA, 91.3 AIME 2026, 83.6 LiveCodeBench v6	1M / open	Together, OpenRouter, local
Qwen3.6-35B-A3B	Efficient open agentic coder (3B active)	86.0 GPQA Diamond, 92.7 AIME 2026, 35B/3B active	262K (to 1M YaRN) / Apache 2.0	Hugging Face, OpenRouter, local
Qwen3.6-27B	Laptop-runnable dense coder	87.8 GPQA Diamond, dense 27B, multimodal	256K / Apache 2.0	Local Mac/PC, Hugging Face, OpenRouter
Rio 3.5 Open 397B	Qwen 3.5 fine-tune, multilingual reasoning	70.8 Terminal-Bench 2.1 (first-party), beats Qwen 3.7 Plus on 4/5	397B/17B active / MIT	Hugging Face, providers, self-host
Llama 4 Maverick	Meta-line flagship	17B active / 400B total params	Llama 4 license	Meta cloud, Hugging Face, local
NVIDIA Nemotron 3 Nano Omni	Edge / low-power	Multimodal, very small footprint	Compact / open	Local, NVIDIA tool

Licensing matters as much as raw score here, and August widened the gap between the two groups. Kimi K2.7 (Modified MIT), DeepSeek V4 (MIT), GLM 5.3 Flash (MIT), MiMo-V2.6-Pro (MIT), GLM-5.2 (MIT), LongCat-2.0 (MIT), Hy3 (Apache 2.0), Hy4 Preview (Apache 2.0), Nex-N2-Pro (Apache 2.0) and Nemotron 3 Ultra (OpenMDW) all clearly allow commercial use with no revenue test.

The rest carry conditions worth reading before you build on them: MiniMax M3 and MiniMax H3 ship under their own community licences, H3 additionally requiring an application from the US, EU, UK and South Korea; Kimi K3 sits on a custom licence with a revenue threshold for anyone reselling it as a service; GLM 5.3 requires a Z.ai security review of any model-as-a-service operator above $10 billion in revenue; and the Qwen Community License 1.0 on Qwen3.8-Flash-Next requires a separate licence from Qwen to run a model-as-a-service or an AI work-assistant business, though internal use is exempt. No open model is first on any Arena board we track any more: Claude Opus 5 took the Agent board in July, the last one an open model led.

How We Evaluate

Benchmarks, Prices, and Hands-On Use

Every ranking on this page combines three inputs: public benchmarks from seven independent houses (Artificial Analysis, Arena formerly LMArena, Scale SEAL, LiveBench, EQ-Bench, ARC Prize and the official Terminal-Bench 4.0 board, covering the Intelligence Index, GPQA Diamond, ARC-AGI-1 through 3, Humanity's Last Exam, GDPval-AA, FrontierMath, HMMT, MCP Atlas, SWE Atlas and the Remote Labor Index), published API and subscription pricing from each vendor's official pricing page, and hands-on use by the FelloAI editorial team running real prompts across the same task on every model. We re-fetch official pricing and benchmark sources before every monthly update.

Benchmarks are weighted to the use case: SWE-bench and Terminal-Bench drive coding, GPQA Diamond and ARC-AGI-2 drive accuracy, GDPval-AA (Artificial Analysis's professional-deliverables benchmark) informs professional-task quality while writing style is judged primarily by hands-on testing, FrontierMath and HMMT drive problem-solving. We disclose when a benchmark is vendor-reported but not independently verified, and we strip any claim we cannot reproduce against a live source. We rank on measured scores: a category crown moves as soon as a model out-scores the holder on the boards that define that category, without waiting for the vote-based boards to catch up, and a model that is not yet widely available is flagged on its card rather than denied one. We re-check ranks as well as figures, because a score can stay correct while the board around it moves. When a model goes through a major upgrade between updates, we re-rank the category and add a "What changed this month" line at the bottom of the deep-dive.

Not sure which model to pick?

Ask ChatGPT, Claude and Gemini and keep the best answer. No three bills.

Fello AI runs on Mac, iPhone and iPad, free to start.

Download Fello AI

Frequently Asked Questions

Common Questions

What is the best AI model right now in October 2026?

It depends on the task. On overall benchmark score, Claude Opus 5.5 (September 22) is #1 on Artificial Analysis's Intelligence Index at 57.6 and Claude Sonnet 5.5 (September 28) is second at 56.0. Opus 5.5 also holds coding, Claude Fable 5.1 holds writing and accuracy, GPT-6 Astra leads hard reasoning, ChatGPT Images 2.5 leads images and MiniMax H3 leads video. For daily chat, GPT-6 has been ChatGPT's default since October 7 and is the assistant most people can actually open. Among open weights, Xiaomi's MiMo-V2.6-Pro leads at 46.3 under plain MIT.

Is GPT-6 out, and is it the best model now?

Yes, it is out, but it is not the best. OpenAI launched GPT-6 Astra on September 3, 2026 and GPT-6 Sol and GPT-6 Luna on September 22, and since October 7 GPT-6 is what ChatGPT answers with: Sol on Plus, Pro, Business and Enterprise, Luna on Free and Go from October 8. Artificial Analysis scores Astra 52.7 on Intelligence Index v4.3.2, behind Claude Opus 5.5 at 57.6, Claude Sonnet 5.5 at 56.0, Meta's Muse Wonder Full at 55.8 and Claude Fable 5.1 at 53.4, though Astra still leads LiveBench Reasoning and ARC-AGI-2. Our GPT-6 Astra breakdown, GPT-6 Sol and Luna breakdown and GPT-6.1 Sol guide have the detail.

Is Gemini 4 out?

Yes. Google announced Gemini 4 Argon on September 30, 2026 and released it on October 13 to Google AI Pro and Ultra subscribers in the Gemini app and to developers on the paid API, at $2 / $10 per 1M tokens until November 30. The free plan and Google AI Plus do not get it. Artificial Analysis scores it 52.6 on Intelligence Index v4.3.2, level with GPT-6 Astra, with the lowest hallucination rate it has measured among leading models, and Arena ranks it first on text and creative writing. Full detail in our Gemini 4 Argon breakdown.

What is Claude Opus 5?

Claude Opus 5 is Anthropic's July 24, 2026 flagship, which Claude Opus 5.5 replaced on September 22; Anthropic now lists it as a legacy model. It scores 50.8 on Intelligence Index v4.3.2 against Opus 5.5's 57.6, and costs $5 / $25 per 1M tokens against Opus 5.5's $4 / $20. ARC Prize confirmed its 30% on ARC-AGI-3, and Arena still ranks it first on Document. Artificial Analysis measures its hallucination rate at 61%, which is why it holds no accuracy crown here.

What is Claude Fable 5.1?

Claude Fable 5.1 is Anthropic's September 1, 2026 release. It scores 53.4 on Intelligence Index v4.3.2, fourth behind Claude Opus 5.5, Claude Sonnet 5.5 and Meta's Muse Wonder Full, but it still has the highest factual accuracy Artificial Analysis has measured, 67%, and the best combined placing on the three prose boards, so it keeps this page's accuracy and writing crowns. It costs $10 / $50 per 1M tokens and shipped alongside the invitation-only Claude Mythos 5.1. Our Claude Fable 5.1 and Mythos 5.1 breakdown has the rest.

Is Claude Fable 5 back?

Yes. Anthropic redeployed Claude Fable 5 on July 1, 2026 after the US government lifted the export-control restriction it had imposed on June 12, and made it permanent in the paid plans on July 20. Max and Team Premium include it at roughly 50% of regular usage limits; Pro and Team Standard reach it through usage credits. API pricing is $10 / $50 per million tokens.

What is GPT-5.6 and can I use it?

GPT-5.6 is OpenAI's previous model family, generally available since July 9, 2026, in three tiers: Luna, Terra and Sol. It is no longer ChatGPT's default, because Chat moved to GPT-6 on October 7, but paid users who picked GPT-5.6 Sol explicitly stay on it, and OpenAI still sells the 5.6 tiers in the API. OpenAI reused the Sol and Luna names for GPT-6, so an unqualified Sol or Luna is now ambiguous.

Is Grok 4.7 out yet?

Yes. SpaceXAI released Grok 4.7 on September 21, 2026 on a new and larger base model than Grok 4.6, and since October 3 it powers Expert mode on grok.com, plus Fast mode for paid subscribers. Artificial Analysis scores it 46.3 on Intelligence Index v4.3.2 at high effort against Grok 4.6's 44.3. API pricing is unchanged at $2 / $6 per 1M tokens, but an index task costs $2.73 against $1.86, because 4.7 writes roughly twice the output. Full numbers are in our Grok 4.7 breakdown.

Which AI is the best for coding?

Claude Opus 5.5. Released September 22, it leads Arena's WebDev board at 1813, ahead of GPT-6 Astra at 1786, at $4 / $20 per 1M tokens. Claude Sonnet 5.5 is the value pick: at $2 / $10 it leads Terminal-Bench 4.0 at 63.6% and LiveBench Coding at 91.4. Claude Haiku 5.5 ($0.10 / $0.50) is the cheap Claude for high-volume sub-agent work, and among open weights GLM 5.3 Flash is the one most teams can host, under plain MIT.

Which AI is the best for writing?

Claude Fable 5.1, because it has the best combined placing across the three prose boards: second on EQ-Bench Creative Writing v3, third on LiveBench Language and ninth on Arena's creative-writing board. Claude Opus 5.5 is the pick if you trust Arena's human votes most, where it is second. Claude Sonnet 5.5 is what most people should actually use, since it is free on claude.ai. Gemini 4 Argon leads Arena's creative-writing board and is on Google AI Pro since October 13, but EQ-Bench and LiveBench have not rated it.

What is the best open-weight AI model in 2026?

Xiaomi's MiMo-V2.6-Pro, published on September 21, 2026 under plain MIT. Artificial Analysis measures it at 46.3 on Intelligence Index v4.3.2, the highest open score it has published, and it runs an index task for $0.13. At 1.02 trillion parameters it needs a cluster, though. For teams that want to host their own, GLM 5.3 Flash (Z.ai, MIT) is the practical pick at 41.8, with 320B total and 18B active parameters.

What is the cheapest frontier-class AI model?

On API price per million tokens, GPT-6 Luna and Claude Haiku 5.5 share the cheapest closed frontier-class rate, $0.10 / $0.50, and Artificial Analysis scores them 38.1 and 43.4. Measured per Intelligence Index task, GPT-6 Luna is the cheapest model on this page at $0.07, ahead of MiMo-V2.6-Pro at $0.13 and Haiku 5.5 at $0.21. Among open weights, GLM 5.3 Flash is our price-performance pick at $0.15 / $0.50 and 41.8 under plain MIT, with DeepSeek V4.1-Flash close behind at $0.15 / $0.60 off-peak.

Which AI models are free?

ChatGPT Free and Go run GPT-6 Luna, rolling out since October 8, under usage limits. Claude Free runs Claude Sonnet 5.5, and free users can also pick Claude Haiku 5.5. The free Gemini app runs on Auto since October 9, which sends most prompts to Flash-Lite, and Gemini 3.6 Flash stays free in Google AI Studio. DeepSeek is free on its chat site, and Grok has a limited free plan. Many models are also open-weight and free to self-host, including MiMo-V2.6-Pro, GLM 5.3 Flash and DeepSeek V4, though the licences differ.

What is Gemini Spark and which plan do you need?

Gemini Spark is Google's always-on cloud AI agent, launched at Google I/O on May 19, 2026. It runs on a Google Cloud VM and works with Gmail, Docs and other Google Workspace apps. Since July 30, 2026 it is included in the $19.99/month Google AI Pro plan, in the US and more than 160 other countries, as well as Google AI Ultra; Google AI Plus, the free tier and work or school accounts do not include it.

What is Fello AI?

Fello AI is an AI chat app for Mac, iPhone and iPad that saves you choosing between AI subscriptions: you ask ChatGPT, Claude, Gemini, Grok and DeepSeek from one app, and new models usually arrive as soon as their API opens. It is free to start, paid plans start at $11.99 a month in most countries, and it is rated 4.7 stars across 27,000+ reviews.

How often do you update this page?

We update this page at least monthly and within 24-48 hours of any major model launch.