Google releases Gemini 3.6 Flash with 17% output token reduction
Google DeepMind released Gemini 3.6 Flash on July 21 at $1.50/$7.50 per 1M tokens, with 17% fewer output tokens than 3.5 Flash.
Google DeepMind released Gemini 3.6 Flash on July 21, the latest workhorse model in the Gemini 3 series. Built on Gemini 3.5 Flash, the model targets coding, knowledge work, and multimodal tasks.
Artificial Analysis Index testing shows that Gemini 3.6 Flash uses 17% fewer output tokens on average than 3.5 Flash, with output tokens dropping from 276,000 to 97,000 on the DeepSWE coding benchmark, a 65% reduction. The model also requires fewer reasoning steps and tool calls, significantly lowering per-task cost.
Pricing is set at $1.50 per 1M input tokens and $7.50 per 1M output tokens, down 17% from 3.5 Flash's $9 output price. Cached input is $0.15 per 1M tokens. The model features a 1M token context window and 64K token output limit.
DeepSWE coding accuracy improved from 37% to 49%, OSWorld-Verified computer use from 78.4% to 83.0%, and MLE-Bench from 49.7% to 63.9%.