Join waitlist
Back to products
Gemini 2.5 Flash-Lite

Gemini 2.5 Flash-Lite

Google's fastest, most cost-efficient model

Website developers.googleblog.com
Overview

What it is

Gemini 2.5 Flash, is now in preview, offering improved reasoning while prioritizing speed and cost efficiency for developers.

Intent

I need it when

Build cost-efficient AI applications with strong performance metrics

Gemini 2.5 Flash offers the best price-to-performance ratio among thinking models and performs comparably to leading models for a fraction of the cost and size. It ranks second only to 2.5 Pro on Hard Prompts in LMArena, making it ideal for developers prioritizing cost efficiency.

Quickly prototype and test AI features with flexible API access

Gemini 2.5 Flash is available in preview via Gemini API, Google AI Studio, Vertex AI, and the Gemini app. Developers can experiment with the thinking_budget parameter through code examples and interactive interfaces without waiting for general availability.

Fine-tune reasoning depth based on task complexity without manual intervention

The model automatically decides how much to think based on perceived task complexity. Developers can set a thinking budget cap, but the model uses only what is needed for the prompt, enabling automatic optimization of reasoning effort.

Solve complex multi-step reasoning problems while minimizing cost and latency

Gemini 2.5 Flash is a hybrid reasoning model that lets developers toggle thinking on/off and set thinking budgets (0-24576 tokens) to balance quality, cost, and speed. Developers can maintain fast speeds of 2.0 Flash while improving performance on complex tasks like math problems and research analysis.

Drop

Not a fit when

  • User requires a fully production-ready model; Gemini 2.5 Flash is currently in early preview and not yet generally available for full production use
  • User needs guaranteed low-latency responses; thinking mode can increase latency depending on reasoning budget settings
  • User requires offline or on-device inference; Gemini 2.5 Flash is accessed via API only
  • User needs transparent pricing information; specific per-token costs are not disclosed in available documentation
  • User requires a model without reasoning capabilities; thinking is a core feature that cannot be fully disabled
  • User needs support for non-English languages exclusively; documentation focuses on English examples
Commercials

Pricing

Pay-as-you-go API pricing; free preview access available