Woofun AI reports that Google has released Gemini 3.6 Flash, a model optimized for programming, multimodality, and multi-step Agent workflows. The update reduces inference steps, tool invocations, and execution loops, resulting in 17% lower output Token usage compared to Gemini 3.5 Flash. API input pricing remains at $1.5 per million Tokens, while output costs drop from $9 to $7.5.

Performance metrics show significant gains, with DeepSWE scores rising from 37% to 49%, MLE Bench from 49.7% to 63.9%, and OSWorld-Verified from 78.4% to 83%. The model supports contexts up to 1 million Tokens and maximum outputs of 64,000 Tokens, enabling Agents to execute tasks more efficiently.

Additionally, Google confirmed that Gemini 3.5 Pro is undergoing partner testing and that Gemini 4 has commenced its most ambitious pre-training round to date, though no release date has been disclosed.