gemini-3.7-flash · Released 13 August 2026
| Provider | |
|---|---|
| Family | Gemini 3 |
| Modality | multimodal |
| Context | 1,000,000 tokens |
| Max output | 64,000 tokens |
| Reasoning | Yes |
| Tool calling | Yes |
| Structured output | Yes |
| Open weights | No / not recorded |
| Licence | — |
No verified Cybatar-run benchmark result has been published for this model yet.
Versioned observations tied to the exact deployment and source date.
Scheduled standard pricing from January 1, 2027.
Evidence ↗Introductory pricing through December 31, 2026.
Evidence ↗Official-source release record for the exact Gemini 3.7 Flash model version.
Evidence ↗Gemini 3.7 Flash launches as a production model for coding and agentic workflows.
Evidence ↗Reported by independent evaluators or the model provider. These are contextual research observations, not Cybatar-run scores.
| Benchmark | Score | Configuration / harness | Provenance | Source |
|---|---|---|---|---|
| Artificial Analysis Intelligence Index v4.1.1 | 56 | Model evaluation · thinking: high | Independent | Artificial Analysis |
| AA-Briefcase 2026-08 | 1131 Elo | Model evaluation · thinking: high | Independent | Artificial Analysis |
| Artificial Analysis Coding Agent Index v1.3 | 57 | OpenCode · thinking: high | Independent | Artificial Analysis |
| DeepSWE 1.1 | 57 % | OpenCode · thinking: high | Independent | Artificial Analysis |
| Terminal-Bench 2.1 | 83 % | OpenCode · thinking: high | Independent | Artificial Analysis |
| SWE-Atlas-QnA 2026 | 31 % | OpenCode · thinking: high | Independent | Artificial Analysis |
| GDPval-AA v2 | 1525 Elo | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| FrontierCode 1.1 Main 1.1 | 43.6 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| DeepSWE 1.1 | 65.3 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| Code Arena 2026-08 | 1588 Elo | Model evaluation · | Provider Reported | Google DeepMind |
| Terminal-Bench 2.1 | 85.8 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| AutomationBench 2026 | 30.4 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| Harvey LAB-AA 2026 | 90.7 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| GDP.pdf 2026 | 34 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| MRCR v2 — 128k v2 | 97 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| OSWorld 2.0 | 47.9 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| Agent's Last Exam 2026 | 26.3 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |
| HLE-Verified 2026-08 | 53.6 % | Model evaluation · thinking: high | Provider Reported | Google DeepMind |