Google Gemini 3.1 Flash-Lite launched: How it is different previous model

Google has launched its newest AI model, Gemini 3.1 Flash-Lite, designed to be the fastest and most cost-efficient in the Gemini 3 series models. Google says that this new version is built for developers who need to process massive amounts of data in a cost-effective manner. The model is rolling out in preview for developers using Google AI Studio and for businesses via Vertex AI.

What 'Flash-Lite' is different

Google explains that the main focus of this new model is balancing extreme speed with deep intelligence. While “Lite” models in the past were often seen as “watered-down” versions of AI, Google claims 3.1 Flash-Lite actually outperforms its predecessors in several key areas.

As per the company, it is 2.5x faster at giving its first answer and has a 45% boost in overall typing speed compared to the older Gemini 2.5 Flash. Moreover, Google has priced this model at $0.25 per million input tokens, making it one of the most cost-effective high-end models on the market. Adding to that, despite being a “Lite” model, it scored a 1432 on the Arena.ai leaderboard, even beating some of the larger models from previous generations in reasoning and understanding.

“Google Gemini 3.1 Flash-Lite achieves an impressive Elo score of 1432 on the Arena.ai Leaderboard and outperforms other models of similar tier across reasoning and multimodal understanding benchmarks, including 86.9% on GPQA Diamond and 76.8% on MMMU Pro–even surpassing larger Gemini models from prior generations like 2.5 Flash,” the company said.

Google Gemini Gemini 3.1 Flash-Lite brings ‘Adaptive Intelligence’

One of the most unique features arriving with this launch is thinking levels which gives developers a “slider” to control how much the AI actually thinks before it speaks. For example, for simple tasks like translating a document or moderating comments, developers can set it to “Low Thinking” to save time and money. For complex tasks, you can dial up the “Thinking” to get deeper, more precise reasoning.

“Early testers highlighted 3.1 Flash-Lite’s efficiency and reasoning capabilities, saying it can handle complex inputs with the precision of a larger-tier model, plus follow instructions and maintain adherence,” Google said.