gemini-3.8-flash is live, and APIYI has it open for calls first
Google’s newest Flash, out September 2, roughly three weeks after 3.7 Flash. Neither Google’s model docs nor its launch blog lists this version yet, so there is no authoritative source for official benchmarks or context specs — we neither relay nor guess at them.
Pricing matches gemini-3.7-flash line for line: $0.75 in / $3.75 out per 1M tokens, $0.0750 cached reads. Migrating from 3.7 means changing the model name and nothing else — request shape, parameters and response fields are unchanged, at no change in cost.
Precisely because the official specs are missing, we made the pre-launch check a thorough one: a paired comparison against 3.7 Flash with both models fired simultaneously on every case — 150 case logs and 198 calls across the Gemini native and OpenAI-compatible protocols. Core capability, reasoning, parallel tool calling, and image and video understanding all line up; response field sets differ in one group with zero type changes, and no regression unique to 3.8 turned up.
Both the default and svip groups are open, as are the OpenAI-compatible and Gemini native endpoints. If you lean heavily on long context, running a slice of traffic first and waiting for the official specs before a full cutover is the safer call.
📖 Full test data and migration notes: Gemini 3.8 Flash Is Live: API Access Ahead of the Docs
← Back to Live Updates · 📚 Monthly archive