Gemini 2.5: Updates to our family of thinking models

Today we are excited to share comprehensive updates to our Gemini 2.5 model range:

• Gemini 2.5 Pro is available and generally stable (no changes from 06-05 preview)

• Gemini 2.5 Flash is generally available and stable (no changes from 05-20 preview, see pricing updates below)

• Gemini 2.5 Flash-Lite is now available for preview

Gemini 2.5 models are thinking models, able to think through their thoughts before responding, resulting in improved performance and improved accuracy. Each model has control over the thinking budget, giving developers the ability to choose when and how much the model “thinks” before creating the response.

Introducing Gemini 2.5 Flash Lite

Today, we’re offering 2.5 Flash-Lite in preview with the lowest latency and cost in the 2.5 model family. It is designed as a cost-effective upgrade from previous Flash 1.5 and 2.0 models. It also provides better performance across most evaluations, lower time to first token while also achieving higher tokens per second decryption. This model is great for high-throughput tasks like large-scale classification or summarization.

Gemini 2.5 Flash-Lite is a logic model that allows dynamic control of the thinking budget using an API parameter. Because Flash-Lite is optimized for cost and speed, “thinking” is turned off by default, unlike our other models. 2.5 Flash-Lite also supports all our native tools such as grounding with Google search, code execution, URL context as well as function calling.

Gemini 2.5 Flash updates and pricing

Over the past year, our research teams have continued to push the Pareto frontier with our Flash model series. When 2.5 Flash was initially announced, we weren’t quite done with the capabilities of 2.5 Flash-Lite. We also launched the “Thinking Price” and the “No-Thinking Price,” which has confused developers.

With the release of the stable release of Gemini 2.5 Flash (the same preview of the 05-20 model we made available at Google I/O), and the incredible performance of 2.5 Flash, we’re updating 2.5 Flash pricing:

• $0.30 / million input tokens (*up from $0.15)

• $2.50 / 1M tokens for output (*less than $3.50 for output)

• We have removed the price difference between thinking and not thinking

• We have kept one price layer regardless of the size of the input symbol

While we strive to maintain consistent pricing between beta and stable releases to minimize disruption, this is a specific modification that reflects the exceptional value of Flash, and still offers the best cost per intelligence available.

With Gemini 2.5 Flash-Lite, we now have a lower cost option (with or without thought) for cost- and latency-sensitive use cases that require less intelligence in the model.

If you are using Gemini 2.5 Flash Preview 04-17, the current preview price will remain in effect until the planned deprecation on July 15, 2025, at which point the model endpoint will be retired. You can move to the generally available “gemini-2.5-flash” model, or switch to the 2.5 Flash-Lite Preview as a less expensive option.

The continued growth of Gemini 2.5 Pro

Growth and demand for the Gemini 2.5 Pro continues to be the steepest of any of our models we’ve ever seen. To allow more customers to build on this model into production, we are making the 06-05 version of the model stable, at the same Pareto frontier price point as before.

We expect that the situations where you need the most intelligence and the most capabilities are where you’ll see the Pro shine, such as programming and agent tasks. Gemini 2.5 Pro sits at the heart of many of the most popular developer tools.

If you are using 2.5 Pro Preview 05-06, the model will remain available until June 19, 2025 and then it will be discontinued. If you are using 2.5 Pro Preview 06-05, you can simply update your model string to “gemini-2.5-pro”.

We can’t wait to see more industries benefit from Pro 2.5 intelligence and look forward to sharing more about expanding beyond Pro in the near future.