Google DeepMind rolled out Gemini 3.7 Flash Hybrid Thinking across Google AI Studio and Vertex AI. The model dynamically balances standard token prediction with chain-of-thought exploration depending on task complexity, pricing input tokens at $0.075/1M to power cost-efficient coding agents.
Key Takeaways
- βDynamically allocates reasoning tokens based on task complexity without manual prompt hacks.
- βInput pricing drops to $0.075 per million tokens for high-volume agent loops.
- βIntegrated with Vertex AI SDK and Google Antigravity framework with 1M context support.