Google · AI Model

Gemini 3.7 Flash

gemini-3.7-flash · Released 13 August 2026

Model profile

ProviderGoogle
FamilyGemini 3
Modalitymultimodal
Context1,000,000 tokens
Max output64,000 tokens
ReasoningYes
Tool callingYes
Structured outputYes
Open weightsNo / not recorded
Licence

Published Cybatar benchmarks

No verified Cybatar-run benchmark result has been published for this model yet.

Model economics

Pricing history

Versioned observations tied to the exact deployment and source date.

All model pricing →
EffectiveDeploymentModeInput / 1MCached / 1MOutput / 1MSource
01 Jan 2027Google standard USD 1.5USD 0.15USD 7.5Source ↗
13 Aug 2026Google standard USD 0.75USD 0.075USD 3.75Source ↗
Lifecycle intelligence

Release & availability timeline

01 Jan 2027 · Pricing Change

Gemini 3.7 Flash pricing observation

Scheduled standard pricing from January 1, 2027.

Evidence ↗
13 Aug 2026 · Pricing Change

Gemini 3.7 Flash pricing observation

Introductory pricing through December 31, 2026.

Evidence ↗
13 Aug 2026 · Model Release

Gemini 3.7 Flash release record

Official-source release record for the exact Gemini 3.7 Flash model version.

Evidence ↗
13 Aug 2026 · Model Release

Google releases Gemini 3.7 Flash GA

Gemini 3.7 Flash launches as a production model for coding and agentic workflows.

Evidence ↗
External research evidence

Published benchmark observations

Reported by independent evaluators or the model provider. These are contextual research observations, not Cybatar-run scores.

BenchmarkScoreConfiguration / harnessProvenanceSource
Artificial Analysis Intelligence Index v4.1.156 Model evaluation · thinking: highIndependentArtificial Analysis
AA-Briefcase 2026-081131 EloModel evaluation · thinking: highIndependentArtificial Analysis
Artificial Analysis Coding Agent Index v1.357 OpenCode · thinking: highIndependentArtificial Analysis
DeepSWE 1.157 %OpenCode · thinking: highIndependentArtificial Analysis
Terminal-Bench 2.183 %OpenCode · thinking: highIndependentArtificial Analysis
SWE-Atlas-QnA 202631 %OpenCode · thinking: highIndependentArtificial Analysis
GDPval-AA v21525 EloModel evaluation · thinking: highProvider ReportedGoogle DeepMind
FrontierCode 1.1 Main 1.143.6 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
DeepSWE 1.165.3 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
Code Arena 2026-081588 EloModel evaluation · Provider ReportedGoogle DeepMind
Terminal-Bench 2.185.8 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
AutomationBench 202630.4 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
Harvey LAB-AA 202690.7 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
GDP.pdf 202634 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
MRCR v2 — 128k v297 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
OSWorld 2.047.9 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
Agent's Last Exam 202626.3 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind
HLE-Verified 2026-0853.6 %Model evaluation · thinking: highProvider ReportedGoogle DeepMind