Overview
Google released Gemini 3.7 Flash on August 13, 2026, positioning it as the primary agentic workhorse of the Gemini 3 family. <cite index="11-11">The release arrived just three weeks after Gemini 3.6 Flash, an unusually short turnaround that Google attributes to developer feedback and algorithmic improvements.</cite> <cite index="13-1,13-2">Google describes the model as "the high-efficiency, cost-effective powerhouse of the Gemini 3 family," delivering Pro-level agentic capabilities and major leaps in code generation and terminal execution.</cite>
Pricing
<cite index="11-12">Through the end of 2026, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens.</cite> <cite index="21-4,21-5">The introductory price expires on December 31, 2026; starting January 1, 2027, standard rates of $1.50 per million input tokens and $7.50 per million output tokens will apply.</cite> Google's own pricing page confirms that <cite index="1-1">Gemini 3.7 Flash and Gemini 3.6 Flash share the same introductory tier of $0.75/$3.75 per one million tokens input/output through December 31, 2026.</cite> The introductory input rate therefore represents a 50% reduction from the standard price that would otherwise apply.
Benchmark Performance
<cite index="23-1">Google published five scores for Gemini 3.7 Flash: FrontierCode 1.1 at 43.6%, DeepSWE v1.1 at 65.3%, WebDev Arena at 1,588 Elo, GDP.pdf at 34.0%, and AutomationBench at 30.4%.</cite> All figures represent gains over the 3.6 Flash baseline. <cite index="18-5,18-6">DeepSWE v1.1 improved from 49.0% to 65.3%, while FrontierCode 1.1 moved from 34.4% to 43.6%, AutomationBench from 17.0% to 30.4%, and GDP.pdf from 22.0% to 34.0%.</cite> Google credits the gains to algorithmic changes rather than a new base model. <cite index="23-2,23-3">As of publication, no independent third party had released a same-generation head-to-head comparison, and no independent SWE-bench Verified comparison against rival models is yet available.</cite>
<cite index="11-3,11-4,11-5">On Terminal-bench 2.1, Gemini 3.7 Flash scores 85.8%, compared with 87.4% for GPT-5.6 Terra. Terra also leads on Terminal-bench 3.0 and OSWorld-2.0, and Claude Sonnet 5 leads on Agent's Last Exam multimodal desktop and operating-system tasks with a 33.3% pass rate, versus 26.3% for Gemini 3.7 Flash.</cite>
Agentic and Enterprise Capabilities
<cite index="13-5">Gemini 3.7 Flash serves as the primary agentic workhorse in the Gemini 3 family, bridging the gap between deep-reasoning Pro models and high-throughput Flash-Lite models while delivering high token efficiency and multi-step multimodal processing.</cite> <cite index="19-6">Google says the model is better at adapting when it hits a roadblock, clarifying intent when a request is ambiguous, and following instructions with greater fidelity—behaviors described as critical for long agent runs.</cite>
<cite index="26-4,26-5,26-6">For knowledge-dense fields such as finance, law, and biosciences, 3.7 Flash delivers improved reasoning and accuracy, significantly outperforming 3.6 Flash on the GDP.pdf benchmark (34.0% vs. 22.0%) and on AutomationBench (30.4% vs. 17.0%).</cite>
Access and Distribution
<cite index="17-3,17-4,17-5">Developers can explore agent-first workflows in Google Antigravity or build via the Gemini application programming interface (API) through Google AI Studio and Android Studio, while enterprise customers access Gemini 3.7 Flash through the Gemini Enterprise Agent Platform.</cite> <cite index="18-14">The model's context window is 1 million tokens in and 64,000 tokens out.</cite>
Competitive Context
<cite index="11-6,11-7">Google's own results do not show 3.7 Flash universally displacing higher-priced competitors; they instead suggest a model that has become substantially more competitive in coding and agent workloads while occupying a lower price tier.</cite> <cite index="18-7">Analysts note the 50% introductory cut is hard-coded to expire January 1, 2027, doubling costs for any team that scales into it during the promotional window.</cite> The release continues a broader pattern of aggressive Flash-tier pricing across the generative AI sector, with Google, Anthropic, and others competing intensely for enterprise application programming interface (API) volume at scale.