Google ships Gemini 3.7 Flash as its coding workhorse while flagship stays delayed

Google released Gemini 3.7 Flash on August 13, 2026, calling it the company’s most intelligent workhorse model yet for coding and AI agents, three weeks after it shipped Gemini 3.6 Flash. The fast cadence lands while Google’s flagship Gemini 3.5 Pro remains delayed with no release date, leaving Flash to carry the company’s near-term push in developer and agent workloads.

Google says the new model posts sharp benchmark gains over its predecessor. On DeepSWE v1.1, a long-horizon software-engineering test, it reports a jump to 65.3 percent from 49.0 percent, and on the WebDev Arena leaderboard its Elo rating rose to 1,588 from 1,538. Document-comprehension and business-automation scores roughly doubled, according to the company’s launch post. None of the figures are independently verified.

Pricing undercuts the prior generation. Introductory rates run $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026 — half the 3.6 Flash launch price — before standard pricing of $1.50 and $7.50 takes effect January 1, 2027.

Gemini 3.7 Flash is available in Google AI Studio, Android Studio, Google Antigravity and the Gemini Enterprise Agent Platform, and is rolling out in Gemini Spark for Google AI Pro and Ultra subscribers across more than 160 countries.

The vendor-reported benchmarks describe controlled tests, not production behavior, and say nothing about how the model handles messy real-world code or long agent runs. The more telling signal is cadence: Google is iterating Flash on a three-week clock while the Pro flagship stays dark, a sequencing that puts its cheapest tier at the center of the roadmap.