Just three weeks after Gemini 3.6 Flash, Google shipped Gemini 3.7 Flash on August 13 with coding upgrades and aggressive pricing. The model scores 43.6% on FrontierCode 1.1 Main—ahead of Claude Sonnet 5 (42.7%) and GPT-5.6 Terra (41.3%) in Google's own benchmarks. On DeepSWE v1.1, it jumps from 49.0% to 65.3%, though Terra still leads at 69.6%. Web development shows a strong gain, with an Elo score of 1588 on Code Arena versus 1538 for the prior version.
The price play
Through December 31, 2026, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens—exactly half the 3.6 Flash launch price and roughly one-third the blended cost of Claude Sonnet 5 ($2/$10) or GPT-5.6 Terra ($2/$12). At an 80/20 input-output mix, that translates to $1.35 per million tokens versus $3.60 for Sonnet 5 and $4.00 for Terra. Starting January 1, 2027, the price doubles to $1.50/$7.50.
Google positions 3.7 Flash as a workhorse for high-volume coding, agents, and knowledge work—not a frontier model. Independent benchmarks show GPT-5.6 Terra still leads on computer-use tasks (Terminal-bench, OSWorld), and Claude Sonnet 5 edges ahead on complex agentic reasoning.
The ownership gap
Despite the competitive pricing, Gemini 3.7 Flash remains API and enterprise only. No open weights, no self-hosting option, no air-gapped deployment. You rent the intelligence at Google's posted rate until the intro period ends.