Google has released Gemini 3.6 Flash, and Gemini 4 is already in pre-training. The new model is aimed at coding, knowledge work, and multimodal agents, and it requires fewer reasoning steps and tool calls when completing multi-step tasks.
Artificial Analysis testing shows Gemini 3.6 Flash uses 17% fewer output tokens than Gemini 3.5 Flash. API input pricing stays at $1.50 per million tokens, while output pricing drops from $9 to $7.50 per million tokens. With token usage and unit costs both down, agents running long tasks will cost less.
Performance has also improved. DeepSWE…