Google DeepMind Releases Gemini 3.1 Flash-Lite for High-Volume Cost-Efficient Deployments

Google DeepMind Blog·medium signal

Google DeepMind released Gemini 3.1 Flash-Lite on March 3, 2026, the fastest and most cost-efficient model in the Gemini 3 series targeting enterprise use cases requiring broad-scale deployment over frontier capability. The model fills the sub-Pro tier of the Gemini 3.x family and complements the 1M-token context Gemini 3.1 Pro, giving organizations a cost-optimized inference path for high-throughput agentic pipelines where per-call cost is the primary constraint.