Gemini 3.7 Flash
Google · Released 13 Aug 2026 · gemini-3.7-flash
$0.75
Input price / 1M tokens
#18 of 61, cheapest first
$3.75
Output price / 1M tokens
#21 of 61, cheapest first
1.05M
Context window
#11 of 71
299
Output tokens / second
#2 of 61
65.3%
DeepSWE v1.1
#14 of 17
85.8%
Terminal-Bench 2.1
#9 of 17
43.6%
FrontierCode 1.1
#5 of 6
30.4%
AutomationBench
#6 of 8
Summary
Gemini 3.7 Flash is a proprietary model from Google, released on 13 Aug 2026.
At $0.75 input and $3.75 output per million tokens, it is #18 of 61 on input price, cheapest first.
Its 1.05M-token context window ranks #11 of 71.
Measured output speed is 299 tokens per second, #2 of 61.
High-speed, efficient Flash model for everyday coding, agentic tool use and reliable multi-step execution.
Benchmarks · 15 reported
| Benchmark | Score | Bar | Rank | Source |
|---|---|---|---|---|
| DeepSWE v1.1mini-swe agent harness, high thinking | 65.3% | #14 of 17 | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| Terminal-Bench 2.1Terminus 2 harness | 85.8% | #9 of 17 | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| Terminal-Bench 3.0mini-swe agent harness | 14.9% | – | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| FrontierCode 1.1official public leaderboard | 43.6% | #5 of 6 | FrontierCode leaderboard, cited in Google's evaluation report | |
| Code Arena (WebDev)official public WebDev leaderboard | 1,588 | Elo-style | – | Code Arena leaderboard, cited in Google's evaluation report |
| AutomationBenchprivate set, official leaderboard | 30.4% | #6 of 8 | AutomationBench leaderboard, cited in Google's evaluation report | |
| GDPVal-AA v2Artificial Analysis leaderboard, Aug 2026 | 1,525 | Elo-style | – | Artificial Analysis, cited in Google's evaluation report |
| Harvey LAB-AAArtificial Analysis leaderboard | 90.7% | – | Artificial Analysis, cited in Google's evaluation report | |
| CharXiv Reasoningno tools (88.7% with search and code execution) | 84.5% | #3 of 6 | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| LVBenchno tools, 1024 frames | 85.4% | – | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| MRCR v2 (8-needle)128k average | 97% | #1 of 13 | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| OSWorld 2.0partial score, max over 3 runs, single attempt per run | 47.9% | #9 of 10 | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| Agent's Last Exampass rate, ALE-Claw harness | 26.3% | – | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| HLE-Verifiedfull 1,811-item verified set | 53.6% | – | Google Gemini 3.7 Flash evaluation reportSelf-reported | |
| LABBench2terminal with bioinformatics tools and internet | 82.1% | – | Google Gemini 3.7 Flash evaluation reportSelf-reported |
Self-reported means the lab ran the test itself. A rank appears only where several labs report the same version of a benchmark.
Similar models
Models with the most benchmarks in common with Gemini 3.7 Flash, and the closest scores.
DeepSWE v1.1
#14 of 17Long coding tasks in real repos
- Gemini 3.7 Flash65.3
- Gemini 3.8 Flash73.7
- Gemini 3.6 Flash49
- GPT-5.6 Luna67.2
- GPT-5.6 Terra69.6
- GPT-6 Sol68.8
Terminal-Bench 2.1
#9 of 17Terminal tasks, earlier set
Compare Gemini 3.7 Flash side by side
Pre-filled with the two closest models. Swap any of them.
Sources
- price, cache, batch, free tier: ai.google.dev/gemini-api/docs/pricing, checked 2 Oct 2026
- context, max output, modalities, api id: ai.google.dev/gemini-api/docs/models/gemini-3.7-flash, checked 2 Oct 2026
- release date, GA status, introductory price: ai.google.dev/gemini-api/docs/changelog, checked 2 Oct 2026
- knowledge cutoff: deepmind.google/models/model-cards/gemini-3-7-flash/, checked 2 Oct 2026
- benchmarks: storage.googleapis.com/deepmind-media/gemini/gemini_3-7_flash_model_evaluation.p, checked 2 Oct 2026
- output speed: artificialanalysis.ai/models/gemini-3-7-flash, checked 2 Oct 2026
Building on Gemini 3.7 Flash?
Get the architecture right before the bill arrives.
We size caching, routing and fallbacks for your workload on a free call, and tell you where a cheaper model is good enough.