タグ: LLMs

Meituan has open-sourced LongCat-2.0, a 1.6 trillion-parameter Mixture-of-Experts model with a native 1-million-token context window, released on June 30, 2026 under a permissive MIT license. It is the model that quietly topped global developer charts on OpenRouter for two months under the codename “Owl Alpha”, and Meituan says it is the first trillion-parameter model trained and served end-to-end on Chinese-made chips, with no Nvidia GPUs involved. This article breaks down what LongCat-2.0 actually is, how its unusual architecture keeps a 1.6T model cheap to run, what the vendor-reported benchmarks claim, and how you can try it today. We also flag where the hype outruns the evidence, because the headline numbers […]

GLM is the family of large language models built by Zhipu AI, the Chinese lab now trading as Z.ai. If you typed “what is GLM” expecting the statistics term or the Bosch laser measure, this is the other one: GLM is a direct open-weight rival to ChatGPT, Claude and Gemini, and its current flagship GLM-5.2 scores high enough on independent benchmarks to sit in the same conversation as models that cost six times more to run. The reason GLM matters to you is simple. Every model in the family is open-weight and MIT-licensed, so anyone can download it, run it, or build on it for free, and the paid API […]

Most AI models people call “open source” are not actually open source. They are open-weight, a narrower thing, and the gap between the two decides what you can legally do with a model, how much you can trust it, and whether anyone could ever rebuild it. In October 2024 the Open Source Initiative finally drew a hard line with its first formal Open Source AI Definition, and by 2026 that line separates a handful of fully open models from the dozens that just look open. This guide explains what open source AI really means, why the open-weight versus open-source distinction matters to you, and which of today’s big models fall […]

ByteDance just announced Seedance 2.5, an AI video model that generates a full 30-second clip in a single pass with no stitching, breaking a limit that has held most rivals to roughly 10 to 15 seconds. The reveal came at Volcano Engine’s FORCE conference on June 23, 2026, alongside a refreshed Seedance 2.0 that now outputs native 4K with 10-bit color. This is one of the biggest jumps in AI video length so far, and it lands at a sensitive moment for ByteDance after Hollywood studios forced it to halt Seedance 2.0’s global rollout earlier this year. In this article you’ll get the full breakdown of what Seedance 2.5 does, […]

On June 22, 2026, Tokyo-based Sakana AI launched Sakana Fugu, and it works unlike almost any model you’ve used. Instead of trying to be the smartest single brain in the room, Fugu is a language model trained to command a pool of other LLMs, including copies of itself, then blend their answers into one. Its top tier, Fugu Ultra, scores 73.7 on SWE-Bench Pro, a number Sakana says puts it shoulder-to-shoulder with Anthropic’s now-pulled Fable 5 and Mythos Preview. This article breaks down what Sakana Fugu actually is and how its “model that orchestrates models” idea works in plain English. We’ll check whether the Fable 5 comparison holds up, why […]

ByteDance’s new Seed 2.1 Pro scored 1539 on Code Arena: Frontend, landing at #8 in the world and drawing level with Claude Opus 4.6. That is a remarkable result for a model that is not even publicly available yet, and it puts ByteDance into the same frontend tier as Anthropic and Z.ai for the first time. The number that matters most sits underneath that headline. Seed 2.1 Pro reached the top 10 in five of seven subcategories, and in the areas where it is strongest only a short list of frontier labs scored higher. This article breaks down exactly where the model ranks, why its React and design results stand […]

The best open source AI models just closed most of the remaining gap to the closed frontier. Moonshot published the weights for Kimi K3 on July 27, 2026, and Artificial Analysis scores it 57 on its Intelligence Index against 61 for Claude Opus 5, the highest-scoring closed model on that board. GLM 5.2 posts a 62.1 on SWE-bench Pro, DeepSeek V4 hits 80.6 on SWE-Bench Verified, and Gemma 4 runs on a laptop while scoring 85.2% on MMLU-Pro. This guide ranks the open source AI models worth your time in 2026, what each one is genuinely best at, and how to actually run them without a rack of GPUs. We […]

Demis Hassabis is the co-founder and CEO of Google DeepMind, a 2024 Nobel laureate in Chemistry, and the man whose lab built Gemini, the AI model now woven through Google Search, Android, and soon a Gemini-powered Siri. Born in London in 1976, he was a chess master at 13 and a video-game designer in his teens before he set out to, in his words, “solve intelligence, and then use it to solve everything else.” Few people have moved that ambition closer to reality. This bio covers everything you are searching for about him, his net worth, his wife and family, his education, the 2024 Nobel Prize for AlphaFold, his knighthood, […]

Update, July 28, 2026: GLM 5.2 is no longer the strongest open-weight model. Kimi K3 shipped its weights on July 27, 2026 and outscores it on the Artificial Analysis Intelligence Index by 57 to 51. GLM 5.2 keeps a real advantage on licensing and hardware, and the benchmark section below has been rebased. GLM 5.2 arrived on June 13, 2026 with three headline numbers. It is a 744-billion-parameter Mixture-of-Experts design with a 1-million-token context window, under an MIT license that makes the weights free for anyone to download and run. Chinese AI lab Zhipu AI (operating as Z.ai) shipped it as a coding-first frontier model, and the timing was not […]

Moonshot AI released Kimi K2.7 Code on June 12, 2026, and it is one of the largest open-weight coding models you can download right now, though it has since been eclipsed in scale by Moonshot’s newer Kimi K3 flagship. It packs 1 trillion total parameters, activates 32 billion of them per token, runs a 256K-token context window, and ships with open weights on Hugging Face under a Modified MIT license. It is also part of a wave of Chinese coding models making headlines this month, alongside ByteDance’s Seed 2.1 Pro, which just ranked #8 on Code Arena: Frontend. The headline pitch is efficiency, with Moonshot claiming the model uses about […]