AIBRIEF
Flash is Google's smaller, cheaper Gemini tier. The 3.7 version costs $0.75 per million input tokens, and Google is moving its always on Spark personal agent to it.
Google put Gemini 3.7 Flash in general availability on 13 August 2026, twenty-three days after Gemini 3.6 Flash went GA on 21 July. Memeburn reports the smaller-tier model is priced at $0.75 per million input tokens and $3.75 per million output tokens, roughly half of 3.6 Flash's per-token cost.
Google positions 3.7 Flash as its most intelligent "workhorse" model yet, aimed at coding workflows and AI agents. On Google's own numbers, the model scores 43.6% on FrontierCode 1.1 Main (up from 34.4% on 3.6 Flash) and 65.3% on DeepSWE v1.1 (up from 49.0%). Both benchmarks measure first-pass code accuracy and production-ready code generation, but the deltas are Google's own framing rather than independent third-party results.
The agent story sits in the migration, not the benchmark: Google's always-on Spark personal agent is moving to 3.7 Flash for cross-task work like file consolidation, email drafting, and status updates.
The cadence is the news. A sub-month Flash cycle means any team that shipped on 3.6 has roughly three weeks to re-evaluate before their cheaper-tier baseline moves. Eval suites, prompt caches, and routing logic need a refresh each time the tier jumps.