Gemini 2.5: Updates to our family of thinking models

We are excited today to share updates in all fields with the Gemini 2.5 family family:

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

• GIMINI 2.5 Flash is generally available (there are no changes from 05-20 preview, see pricing ups below)

• Gemini 2.5 Flash-Lite is now available in inspection

Gemini 2.5 models are thinking models, able to think through their ideas before response, which leads to promoting performance and improving accuracy. Each model has the control of the thinking budget, giving developers the ability to choose Matthew and the amount of “thinking” the model before creating a response.

Presenting Gemini 2.5 Flash Litt

Today, we offer 2.5 Flash Litt in a preview with the lowest transition time and cost in the 2.5 -style family. It is designed as an effective cost of 1.5 and 2.0 flash models. It also provides better performance across most EVALS, and a lesser time for the distinctive symbol first while achieving the upper codes per second. This model is great for high productivity tasks such as classification or summary widely.

Gemini 2.5 Flash-Lite is a thinking model, which allows dynamic control in the thinking budget with API. Since Flash-Lite was improved for cost and speed, “thinking” is dependent, unlike our other models. 2.5 Flash-Lite also supports all our original tools such as grounding with Google Search, implementing software instructions and URL, as well as calling the job.

Updates to Gemini 2.5 Flash and Prision

During the past year, our research teams continued to pay Pareto Frontier through our Flash Model series. When 2.5 Flash was initially announced, we have not yet finished the capabilities of 2.5 Flash Litt. We also launched “thinking” and “non -thinking price”, which led to the disturbance of developers.

Through the stable version of Gemini 2.5 Flash Rolling (the same as the Self Form 05-20 that we presented in Google I/O), and the amazing performance of 2.5 Flash, we update prices for 2.5 Flash:

• $ 0.30 / 1 million input codes (*higher than $ 0.15 inputs)

• Output icons of 2.50 / 1 million dollars (*decrease from $ 3.50)

• We removed thought against the difference in the price of non -thinking

• We maintained one -price layer regardless of the size of the input code

While we strive to maintain fixed prices between the inspection and stable versions to reduce the disorder, this is a specific amendment that reflects the exceptional Flash value, and still provides the best cost of everything related to writing.

And with Gemini 2.5 Flash-Lite, we now have a low cost option (with or without thinking) for sensitive use of cost and cumin that requires a lower model intelligence.

If you are using Gemini 2.5 Flash Preview 04-17, the current preview pricing will remain in effect until the plan is stopped on July 15, 2025, and at this point this typical end point will be stopped. You can move to the generally available model “Gemini-2.5-Flash”, or switch to 2.5 Flash-Lite Preview as a less expensive option.

Continuing growth from Gemini 2.5 Pro

The growth and demand on Gemini 2.5 Pro remains the most severe in any of our models that we have seen at all. To allow more customers to build on this model in production, we make a 06-05 version of the model, with the same Pareto Frontier price point as it was before.

We expect the cases where you need the highest intelligence and most of the capabilities are where you will see Pro Shine, such as coding and agent tasks. Gemini 2.5 Pro is located in the heart of many of the most popular developers tools.

If you are using 2.5 Pro Preview 05-06, the model will remain available until June 19, 2025 and will be turned off. If you are using 2.5 Pro Preview 06-05, you can simply update your model chain to “Gemini-2.5-PRO”.

We cannot wait to see more areas that benefit from the intelligence of 2.5 Pro and we look forward to sharing more about the scaling beyond Pro in the near future.