Gemini 3.1 Flash-Lite: Built for intelligence at scale

Google has released Gemini 3.1 Flash-Lite, its newest AI model focused on speed and cost-efficiency, which is now available in preview through the Gemini API and Vertex AI. The release targets developers building high-volume applications by providing a new option designed to handle large-scale workloads without the latency or cost typically associated with more powerful models. This positions the model for adoption in services requiring rapid, real-time responses.

Priced at $0.25 per million input tokens and $1.50 per million output tokens, 3.1 Flash-Lite shows notable performance improvements over its predecessor, 2.5 Flash. According to the Artificial Analysis benchmark, it delivers a 2.5 times faster Time to First Answer Token and a 45% increase in output speed. The model has also achieved an Elo score of 1432 on the Arena.ai Leaderboard and performs well on reasoning benchmarks, including 86.9% on GPQA Diamond and 76.8% on MMMU Pro, surpassing some larger models from prior Gemini generations.

The introduction of 3.1 Flash-Lite affects the market by lowering the barrier for integrating AI features that demand high frequency and low latency. Early use cases from companies like Latitude, Cartwheel, and Whering demonstrate its application in complex problem-solving, from mass-scale translation and content moderation to generating user interfaces and creating business agents for multi-step tasks. By offering a model that balances performance with operational cost, Google is aiming to capture a segment of the developer market that needs scalable intelligence for everyday applications.

Google's strategy with Gemini 3.1 Flash-Lite is to commoditize high-frequency AI tasks by offering a model that aggressively competes on speed and price, directly targeting the developer market segment where operational efficiency is paramount.