Google(@Google):Gemini 3.6 Flash: Our smartest everyday model, built directly from your feedback. We took what you loved about Gemini 3.5 Flash and incorporated your feedback on where we could improve. You asked for greater efficiency, so we built an updated model that’s better at coding, reasoning, and tool calling while being substantially more token efficient. It outperforms Gemini 3.5 Flash on quality and precision, using up to 65% fewer output tokens on benchmarks like DeepSWE by Datacurve, all at a lower cost.

2026.07.21 15:30

Gemini 3.6 Flash: Our smartest everyday model, built directly from your feedback. We took what you loved about Gemini 3.5 Flash and incorporated your feedback on where we could improve. You asked for greater efficiency, so we built an updated model that’s better at coding, reasoning, and tool calling while being substantially more token efficient. It outperforms Gemini 3.5 Flash on quality and precision, using up to 65% fewer output tokens on benchmarks like DeepSWE by Datacurve, all at a lower cost.