Short answer
Third-party pricing pages report Gemini 2.5 Pro (gemini-2.5-pro) at $1.25 per 1 million input tokens and $10.00 per 1 million output tokens in the lower tier, and $2.50 input and $15.00 output per 1 million tokens in the higher tier. Finout labels these reference rates in USD. These are not a confirmed current Google quote for the standard paid Gemini Developer API: the Google pricing information reviewed here does not verify Gemini 2.5 Pro’s specific rates or explain how the reported 200,000-token threshold is applied. (aistudio.google.com) (aistudio.google.com)
The production status notices for the Gemini Developer API conflict. Both Google notices restrict access to people who previously used the 2.5 models. One says Gemini 2.5 Pro is not deprecated and will continue to be served until further notice; the AI Studio deprecation notice says requests will route to Gemini 3.8 Flash and be billed at its rates beginning December 1, 2026. The available notices do not establish which statement supersedes the other. Google Cloud’s separate Gemini Enterprise Agent Platform lists an October 20, 2026 retirement date; that is a different product channel. (aistudio.google.com)
Reported Gemini 2.5 Pro API rates
These are third-party reported reference rates, not a confirmed quote for a particular Google API channel, billing mode, or current deployment:
| Token category | Reported rate per 1 million tokens | Reported tier |
|---|---|---|
| Input | $1.25 USD | At or below 200,000 tokens |
| Output | $10.00 USD | At or below 200,000 tokens |
| Input | $2.50 USD | Above 200,000 tokens |
| Output | $15.00 USD | Above 200,000 tokens |
Finout reports these rates, but the cited material does not establish which token count selects the tier or whether crossing the boundary changes the rate for all applicable tokens. Confirm the current price and tier rules for your specific API use before budgeting. (aistudio.google.com)
Example: a conditional token subtotal
For illustration, assume a request has 100,000 input tokens and 10,000 output tokens counted at the reported output rate:
- Input: 100,000 ÷ 1,000,000 × $1.25 = $0.125
- Output: 10,000 ÷ 1,000,000 × $10.00 = $0.10
- Subtotal: $0.225, or about $0.23
This is arithmetic using Finout’s reported lower-tier rates—not a verified Google charge. It assumes the request qualifies for the lower tier and that 10,000 is the billable output-token count, not merely the visible response length. Finout says thinking tokens are billed alongside visible output, but that third-party statement does not establish the billing rule for every Google channel. (aistudio.google.com)
For a production budget, estimate input and billable output separately using representative usage, then apply rates and tier rules confirmed for your channel. Treat the result as a token subtotal unless you have confirmed any applicable platform fees or other billable features. The Google pricing page describes paid access for production applications and lists a general Batch API cost reduction, but the information reviewed here does not confirm a Gemini 2.5 Pro rate or its eligibility for that discount. (aistudio.google.com)
Production availability and lifecycle
- Gemini Developer API access: Google says access to the 2.5 models is limited to users who have actively used them before. Its notices disagree about whether Gemini 2.5 Pro will continue to be served or will be rerouted on December 1, 2026. Check Google’s current instructions for your account before relying on the model or its reported rates. (
aistudio.google.com) - Google Cloud Gemini Enterprise Agent Platform: Google’s lifecycle documentation lists Gemini 2.5 Pro’s retirement date as October 20, 2026, which has passed as of November 9, 2026. This is separate from the Gemini Developer API notices above. (
docs.cloud.google.com)
Bottom line: The commonly reported reference rates are $1.25/$10 per million input/output tokens in the lower tier and $2.50/$15 above the reported 200,000-token boundary. They are not enough by themselves to produce a confirmed current production quote. Verify the applicable channel, lifecycle instructions, rate, tier selection, and billable usage before budgeting.
Sources
Gemini Pricing in 2026 for Individuals, Orgs & Developers — Finout
Gemini Developer API pricing — Google AI Studio
Gemini deprecations — Google AI Studio
Gemini deprecations — Google AI for Developers
Model versions and lifecycle — Gemini Enterprise Agent Platform