Google has officially announced the release of Gemini 3.1 Flash Lite Preview, the latest offering in the Gemini 3 family. This model notably features a 45% increase in output speed, making it the fastest among its predecessors. Now available via Google AI Studio and Vertex AI, Gemini 3.1 Flash Lite is designed to handle high-volume, cost-sensitive workloads efficiently.
Priced at $0.25 per million input tokens and $1.50 per million output tokens, the new model underscores Google’s commitment to offering competitive price-performance. Available in preview through the Gemini API, Google AI Studio, and Vertex AI, it targets developers with substantial processing needs, offering up to 363 tokens per second.
The Gemini 3.1 Flash Lite model outperforms its predecessor, Gemini 2.5 Flash, across most benchmarks and offers a significant advantage over competitors like GPT-5 mini and Claude 4.5 Haiku. One of its noteworthy features is the capability to adjust reasoning time via the API, adding a layer of cost flexibility for large-scale tasks.
Although no agent benchmarks are published, as the model is primarily designed for data processing rather than agentic orchestration, the Gemini 3.1 Flash Lite promises a marked improvement in quality and performance, making it a valuable tool for developers seeking efficient and scalable solutions.

Collections

Google has released Gemini 3.1 Flash Lite Preview, the newest model in the Gemini 3 family. It delivers a 45% increase in output speed compared to predecessors, reaching up to 363 tokens per second. Priced at $0.25 per million input tokens and $1.50 per million output tokens, it targets high-volume, cost-sensitive workloads. The model is available via the Gemini API, Google AI Studio, and Vertex AI. It outperforms Gemini 2.5 Flash on most benchmarks and claims advantages over GPT-5 mini and Claude 4.5 Haiku. A notable API feature allows adjusting reasoning time for cost flexibility, though no agent benchmarks are provided as the model focuses on data processing rather than agentic tasks.

Google Announces Gemini 3.1 Flash Lite Preview with Enhanced Output