Gemini 3.7 Flash launched with improved coding capabilities and half price

Google has released Gemini 3.7 Flash just three weeks after Gemini 3.6 Flash. Google calls this the most powerful main model in terms of coding and AI agents so far, and attributes the rapid improvement in intelligence level to "excellent algorithm improvements." Coding performance has been greatly improved. Code quality is the most obvious improvement in this upgrade. Gemini 3.7 Flash scored 43.6% on the FrontierCode benchmark, which was higher than the previous generation’s 34.4%; it scored 65.3% on the DeepSWE benchmark, which was higher than the previous generation’s 49.0%. By Google's own measure, both scores were ahead of the Claude Sonnet 5 and GPT-5.6 Terra. Benchmark tests on Google also show improvements in the model's web development, document understanding, and business process automation. Pricing halved until year-end Gemini 3.7 Flash is available via API, AI Studio and Antigravity. The initial pricing is US$0.75 per million input Tokens and US$3.75 per million output Tokens, which is 50% cheaper than when Gemini 3.6 Flash was launched. Both models are currently priced the same. That pricing is maintained until the end of the year, but the two models themselves presumably won't be maintained until then. Source