Gemini 3.8 Flash is GA — same price as 3.7, twenty days later

Google released Gemini 3.8 Flash today. The Gemini API changelog for September 2 reads: "Gemini 3.8 Flash generally available (GA): Released gemini-3.8-flash, our most intelligent Flash model, engineered for long-horizon software engineering, autonomous agents, and complex enterprise workflows."

The numbers, from Google's own pricing page and model metadata: $0.75 in / $3.75 out per million tokens through December 31, 2026, then $1.50 / $7.50 from January 1 — identical to 3.7 Flash, introductory clock and all. Context caching $0.075. Input limit 1,048,576 tokens, output limit 65,536. So the price card did not move; what Google is selling is the positioning — 3.7 was "high-speed, efficient" for everyday coding, 3.8 is aimed at long-horizon engineering and agents.

The cadence is the story. 3.6 Flash shipped July 21, 3.7 on August 13, 3.8 on September 2 — three Flash revisions in six weeks, and Sundar Pichai has said roughly one model a month is the target. The next-Flash hunt on our board is now armed on 3.9.

On our end, honestly: the radar read gemini-3.8-flash at 09:31:44 UTC this morning from two catalogs and filed it as an early variant rather than a launch, because "flash" was a variant marker even for an entry that is itself a Flash model. The alert did not fire; this note is the human catching up. The classifier now treats a token that is part of a model's own name as the model, and a test pins it. Looking back, no Flash-family launch had ever fired automatically — this one is the fix.