Google released Gemini 3.7 Flash on August 13, 2026, just 23 days after Gemini 3.6 Flash. The coding numbers moved a long way in three weeks: 43.6% on FrontierCode 1.1 Main against 34.4% for its predecessor, and 65.3% on DeepSWE v1.1 against 49.0%. It launched at $0.75 per million input tokens and $3.75 per million output tokens.

Two details went almost entirely unreported on launch day, and both change the decision. That headline price is introductory and expires on December 31, 2026, after which it doubles. And Google's own footnote excludes the European Economic Area, the United Kingdom, Switzerland and Nigeria from the only consumer surface the model runs on. This article covers the verified benchmarks, what each one actually measures, the real pricing picture next to cheaper rivals, where you could reach the model at launch, and whether switching was worth the effort.

Update, September 2, 2026: Google has since shipped a successor. Gemini 3.8 Flash carries identical specifications and the same $0.75 and $3.75 rates, and beats this model on every benchmark Google published, so anyone choosing what to call today should start there.

Update, November 12, 2026: Google has since released Gemini 4 Hydrogen, which it calls the recommended upgrade from Gemini 3.8 Flash and earlier Gemini 3 models. It costs $1.50 per million input tokens and $7.50 per million output tokens, twice the rate of this model, and it is the first Gemini 4 model with a free API tier.

Update, November 19, 2026: Google deprecated Gemini 3.7 Flash on October 8, 2026 and replaced it with Gemini 3.8 Flash. The model can no longer be selected in Google AI Studio, and API requests to gemini-3.7-flash are now routed automatically to gemini-3.8-flash. Google says the endpoint is not shutting down, but its price list no longer carries a Gemini 3.7 Flash row. The rest of this article describes the model as it launched.

The Key Takeaways

  • Gemini 3.7 Flash launched August 13, 2026 at $0.75 input / $3.75 output per million tokens, 23 days after Gemini 3.6 Flash. That price was introductory, with $1.50 / $7.50 due from January 1, 2027, exactly what 3.6 Flash cost at launch.
  • Coding is the real upgrade. DeepSWE v1.1 climbs from 49.0% to 65.3% and AutomationBench nearly doubles, from 17.0% to 30.4%.
  • Independent testing backs it up. At launch Artificial Analysis scored it 56 against 52 for 3.6 Flash, and ranked it first of 186 models on output speed at 340.1 tokens per second.
  • Most of Europe was locked out. At launch, consumer access ran through Spark only, and Google excluded the EEA, UK, Switzerland and Nigeria.
  • It was not the cheapest option. GPT-6 Luna costs $0.10 / $0.50 and DeepSeek V4.1-Flash $0.15 / $0.60 off-peak, doubling to $0.30 / $1.20 at peak, all well under its launch rate.
  • Deprecated on October 8, 2026. Google replaced it with Gemini 3.8 Flash, and API requests to gemini-3.7-flash now route to 3.8 Flash automatically.

What Is Gemini 3.7 Flash?

Vom Herausgeber

Finde heraus, welche KI am besten antwortet

Stell GPT, Claude und Gemini dieselbe Frage und vergleiche die Antworten. Auf Mac, iPhone und iPad.

Jetzt herunterladen!
Vom Herausgeber

Finde heraus, welche KI am besten antwortet

Stell GPT, Claude und Gemini dieselbe Frage und vergleiche die Antworten. Auf Mac, iPhone und iPad.

Kostenlos testen
Vom Herausgeber

Schluss mit drei KI-Abos

GPT, Claude und Gemini in einer App und mit einem Abo. Auf Mac, iPhone und iPad.

Jetzt herunterladen!
Vom Herausgeber

Schluss mit drei KI-Abos

GPT, Claude und Gemini in einer App und mit einem Abo. Auf Mac, iPhone und iPad.

Kostenlos testen
Vom Herausgeber

Schluss mit dem Hin und Her zwischen KI-Apps

Stell ChatGPT, Claude und Gemini dieselbe Frage in einem Fenster und nimm die beste Antwort. Auf Mac, iPhone und iPad.

Kostenlos testen
Vom Herausgeber

Schluss mit dem Hin und Her zwischen KI-Apps

Stell ChatGPT, Claude und Gemini dieselbe Frage in einem Fenster und nimm die beste Antwort. Auf Mac, iPhone und iPad.

Jetzt herunterladen!
Vom Herausgeber

Nicht nur über KI-Modelle lesen. Selbst testen

Die neuesten KI-Modelle, startklar in einer App. Auf Mac, iPhone und iPad.

Jetzt herunterladen!
Vom Herausgeber

Nicht nur über KI-Modelle lesen. Selbst testen

Die neuesten KI-Modelle, startklar in einer App. Auf Mac, iPhone und iPad.

Kostenlos testen

Gemini 3.7 Flash was Google's fast, low-cost tier model until Gemini 3.8 Flash replaced it, tuned for coding, agent workflows and multi-step execution rather than raw reasoning. Google's launch announcement calls it "our most intelligent workhorse model yet for coding and agents", and the benchmark selection backs that framing up. Every headline number Google published is a software engineering or automation test.

The cadence is the story as much as the model. Google has now shipped three Flash releases in under two months while Gemini 3.5 Pro, promised for June, still has not appeared. Flash stopped being the budget sideshow some time ago. It is currently Google's entire public shipping rhythm.

The full spec sheet

Most launch-day coverage skipped the specifications entirely, because Google put them on the model card rather than in the announcement. They are worth reading, because one of them is a bigger practical upgrade than any coding benchmark.

SpecificationGemini 3.7 Flash
Model codegemini-3.7-flash
ReleasedAugust 13, 2026
Input context window1,048,576 tokens (1M)
Output limit65,536 tokens (64k)
Knowledge cutoffMarch 2026
Input modalitiesText, image, video, audio, PDF
Thinking levelsLow, medium, high (minimal returns an error)
Input price$0.75 per 1M tokens (introductory, at launch)
Output price$3.75 per 1M tokens (introductory, at launch)
Context caching$0.075 per 1M tokens, plus $0.50 per 1M tokens per hour storage (at launch)
Safety updatesCBRN and cyber offence safeguards

The figures come from the Gemini 3.7 Flash model card, which is the authoritative reference if you are wiring this into production. Two details deserve more attention than they got. The context window and output limit match Gemini 3.6 Flash exactly, so nothing was traded away for the coding gains. And the knowledge cutoff holds at March 2026, though the model card notes some domains reflect data only to January 2025, so treat questions about the last few months with care.

The thinking configuration is the practical lever here. Gemini 3.7 Flash accepts low, medium and high thinking levels and defaults to medium, and the minimal setting that worked on some earlier models now returns an error. Since output tokens include thinking tokens, that setting is the difference between a cheap call and an expensive one.

Gemini 3.7 Flash Pricing

Nearly every launch-day write-up reported the same line: half the price of Gemini 3.6 Flash. That is true against the price 3.6 Flash launched at, and misleading about what 3.6 Flash cost by then.

The expiry date in the footnote

Google's own footnote is unambiguous: "Introductory pricing expires on December 31, 2026. Starting January 1, 2027, $1.50/1M input tokens and $7.50/1M output tokens will apply." That is not a discount on a new lower rate. It is a four-and-a-half month promotion, after which Gemini 3.7 Flash would have cost precisely what Gemini 3.6 Flash cost on its launch day.

There is a second wrinkle the comparison hides. Google also moved Gemini 3.6 Flash onto the same $0.75 and $3.75 introductory rate. At launch the two models cost exactly the same, so the upgrade decision came down to capability, not price. Our full breakdown of Gemini 3.6 Flash covers what the older model still does well.

How it compares to cheaper rivals

No ranking article on launch day put Gemini 3.7 Flash next to a non-Google model. Once you do, the value story gets more complicated.

ModelInput per 1MOutput per 1MNotes
Gemini 3.7 Flash$0.75$3.75Launch rate; deprecated October 8, 2026 and no longer on Google's price list
Gemini 3.6 Flash$0.75$3.75Same introductory rate today
Gemini 3.5 Flash-Lite$0.30$2.50Cheapest Gemini tier
Claude Haiku 5.5$0.10$0.50Prompts up to 100K tokens; replaced Haiku 4.5 ($1.00 / $5.00) on October 7, 2026
GPT-6 Luna$0.10$0.50About a seventh of Google's input rate, halved from $0.20 / $1.20 on September 22, 2026
DeepSeek V4.1-Flash$0.15 / $0.30$0.60 / $1.20Off-peak / peak, cut again on September 10, 2026. GPT-6 Luna now undercuts it on both axes at every hour

Gemini 3.7 Flash was not the budget option. It undercut Claude Haiku 4.5 on both input and output, but GPT-6 Luna costs about a seventh as much on both input and output, after OpenAI halved its cheap tier to $0.10 / $0.50 on September 22, 2026, and Anthropic matched that rate when Claude Haiku 5.5 replaced Haiku 4.5 on October 7, 2026. DeepSeek V4.1-Flash is also well under Google, at a fifth of its input rate off-peak. DeepSeek split into peak and off-peak pricing on August 16, 2026 and cut the Flash rate again on September 10, so V4.1-Flash now runs $0.15 input and $0.60 output off-peak and $0.30 / $1.20 at peak, staying under Gemini 3.7 Flash's launch rate at every hour, though Luna's September cut has since taken OpenAI below DeepSeek on both axes. Anyone picking purely on cost per token has three better answers, and all three get further ahead of Google's Flash models when the promotion lapses in January. Our AI pricing comparison tracks the full field as these rates move.

Gemini 3.7 Flash Benchmarks

Google published five benchmark pairs against Gemini 3.6 Flash. The new model wins every row, and two of the margins are large enough to matter in practice.

BenchmarkWhat it measures3.7 Flash3.6 Flash
FrontierCode 1.1 MainHard competitive programming problems43.6%34.4%
DeepSWE v1.1Long-horizon software engineering65.3%49.0%
WebDev ArenaHuman-voted web development quality (Elo)15881538
GDP.pdfComprehension of complex documents34.0%22.0%
AutomationBenchMulti-step enterprise workflow automation30.4%17.0%

The two that stand out are DeepSWE v1.1 and AutomationBench. A jump from 49.0% to 65.3% on long-horizon software engineering is a 16-point gain on the test closest to what an agent does over a long session. AutomationBench nearly doubling, from 17.0% to 30.4%, points the same direction. Both measure whether a model can hold a plan together across many steps without drifting, which is exactly where cheap models have historically fallen apart.

One caveat belongs next to those numbers. An AutomationBench score of 30.4% still means the model fails roughly seven out of ten multi-step automation tasks. The improvement is large and the absolute number is low, and Google chose to lead with the delta rather than the level.

The model card carries four more scores that did not make the announcement, and they fill in the shape of the model. Gemini 3.7 Flash hits 97.0% on GDM-MRCR v2 at 128k, which is near-perfect long-context retrieval, and 85.8% on Terminal-bench 2.1. Computer use lands at 47.9% on OSWorld-2.0. Then there is Terminal-bench 3.0, where the score drops to 14.9%. That last figure is the honest one. Move to a harder version of the same test and the model solves roughly one task in seven, which is a useful reminder of how much benchmark numbers depend on which edition you quote.

What independent testing found

Vendor benchmarks are vendor benchmarks, so the more useful question is what neutral measurement says. Artificial Analysis has already scored the model, and its results are stronger than Google's own framing suggested.

Measure at launchGemini 3.7 FlashGemini 3.6 Flash
Intelligence Index v4.1.15652
Output speed340.1 tokens/sec210.0 tokens/sec
Blended price per 1M$0.58$0.63 (was $1.16 before Google matched the rate)
Context window1.0M1.0M

Two results deserve attention. At launch, Artificial Analysis ranked Gemini 3.7 Flash first out of 186 models on output speed at 340.1 tokens per second. It also scored 56 on Intelligence Index v4.1.1, ahead of Claude Sonnet 5 at 55 and just behind GPT-5.6 Terra and Muse Spark 1.2 at 57. For the flagship matchup, see how Gemini stacks up against Claude.

That second number is the one that separates this launch from the last one. Gemini 3.6 Flash arrived faster and cheaper than its predecessor while scoring identically on the same index, which earned it a round of underwhelmed coverage. Gemini 3.7 Flash moves from 52 to 56 and launched at half the blended cost at the same time. Google then put 3.6 Flash on the same list price, so the two cost within a few cents of each other and the quality gap was the whole argument. This release is not just cheaper. It is measurably better. For where that puts it against the wider field, our best AI models rankings are updated within 48 hours of any major launch.

Where You Could Use Gemini 3.7 Flash

Developer access was broad from day one. At launch the model went live in the Gemini API through Google AI Studio, in Android Studio, in Google Antigravity, and across the Gemini Enterprise Agent Platform and app. Since its deprecation on October 8, 2026 it can no longer be selected in Google AI Studio, and API requests to it are answered by Gemini 3.8 Flash.

Consumer access was a different story, and much narrower than the headlines implied. In the Gemini app, Gemini 3.7 Flash reached users through Spark only, Google's always-on personal agent, and Spark required a Google AI Pro or Ultra subscription. There was no free tier route to this model in the app. Our comparison of Gemini Spark against Claude Cowork covers what that agent actually does once you have it.

The regions Google left out

The announcement says Spark is available to subscribers "in over 160 countries", and almost every write-up stopped at that number. The footnote continues: the rollout excludes the European Economic Area, Nigeria, Switzerland and the United Kingdom.

For a large share of European readers that single line outranked every benchmark in this article. In the EU, the UK or Switzerland, Gemini 3.7 Flash could not be reached through the Gemini app at all, whatever you paid for AI Pro or Ultra. The API route was open, so developers in those regions were unaffected, and Google gave no timeline for everyone else.

Why Google Keeps Shipping Flash

Gemini 3.5 Pro was announced at I/O in May 2026, where Google said it looked forward to rolling the model out the following month. June came and went. Google has not issued a second date since, and said nothing about Pro alongside this launch. Bloomberg reported in July that the model was held back to improve its coding performance, and that an attempt to fix this by updating training data produced disappointing results.

The newsletter SemiAnalysis wrote in early August that Google had quietly cancelled Gemini 3.5 Pro and was promoting Gemini 4 in its place. Google has not confirmed that, no tier-one outlet has matched it, and Google's own last on-record position, from Logan Kilpatrick in July, was that 3.5 Pro is testing with partners. Treat the cancellation story as one newsletter's claim rather than established fact.

What is verifiable is the pattern. Three Flash releases in seven weeks, a Pro model roughly three months past its only stated window, and a separate announcement that pre-training has begun on Gemini 4. Our running coverage of what is known about Gemini 4 follows where the Pro line goes next.

Should You Have Switched to Gemini 3.7 Flash?

At launch the answer for anyone calling Gemini 3.6 Flash was yes: the two models cost the same, and 3.7 Flash won every published benchmark, several by wide margins. That advice has expired. Google deprecated Gemini 3.7 Flash on October 8, 2026 and routes requests for it to Gemini 3.8 Flash, so if you are moving off 3.6 Flash today, move to Gemini 3.8 Flash, which Google lists at the same introductory rate.

If you are choosing a budget model from scratch for agentic coding, test Gemini 3.8 Flash rather than this model: it scores 73.7% on DeepSWE v1.1 against 65.3% here. If you are choosing on price alone, pick GPT-6 Luna or DeepSeek V4.1-Flash instead, and remember that Google's Flash rate doubles in January while theirs do not.

The broader point is that no single model wins across the board any more, and the cheap tier now changes every three weeks. Running one subscription against one provider means re-evaluating every time a launch like this lands. Fello AI puts Gemini on your Mac alongside ChatGPT, Claude, Grok and DeepSeek in a single native app, so comparing a new release against the model you already use takes one click rather than another API key.

Frequently Asked Questions

Is Gemini 3.7 Flash free?

It was not free in the Gemini app, where it ran only inside Spark on a Google AI Pro or Ultra subscription, though the Gemini API free tier listed the model at launch. Since October 8, 2026 the model is deprecated: it can no longer be selected in Google AI Studio, and API requests to gemini-3.7-flash are routed to Gemini 3.8 Flash.

What is the Gemini 3.7 Flash context window?

1,048,576 tokens, or 1M, with a 65,536-token output limit. Those figures match Gemini 3.6 Flash exactly. They appear on the Gemini 3.7 Flash model card and in the Gemini API model documentation rather than in the launch announcement, which is why most launch-day coverage omitted them.

Why couldn't I use Gemini 3.7 Flash in Europe?

Spark, the only consumer surface running the model at launch, excluded the European Economic Area, the United Kingdom, Switzerland and Nigeria. Google stated the exclusion in a footnote to the launch post and gave no timeline for those regions. API access was unaffected until October 8, 2026, when Google deprecated the model and began routing its API requests to Gemini 3.8 Flash.

When does the Gemini 3.7 Flash introductory price end?

Google set it to end on December 31, 2026, with the rate rising from $0.75 to $1.50 per million input tokens and from $3.75 to $7.50 per million output tokens on January 1, 2027, and context caching doubling from $0.075 to $0.15 per million tokens. Since the deprecation on October 8, 2026, Google's price list no longer carries a Gemini 3.7 Flash row, and it has not said how requests routed to Gemini 3.8 Flash are billed.

Is Gemini 3.7 Flash better than Gemini 3.5 Pro?

There is no way to compare them, because Gemini 3.5 Pro has never shipped. It was announced at I/O in May 2026 for release the following month, and Google has issued no new date since. Bloomberg reported coding performance as the cause of the delay. Google gave no Pro release date alongside this launch either.

Das Titelbild dieses Artikels wurde mithilfe von KI erstellt.