Gemini 2.5 API prices at a glance
The commonly reported Gemini 2.5 API rates are $0.10–$2.50 per 1 million input tokens and $0.40–$15 per 1 million output tokens, depending on the model and, for Pro, prompt length. These are third-party-reported reference rates—not a verified current Google price quote. The Pro and Flash figures below come from eesel’s Gemini API pricing table, which frames its API coverage as developer API pricing; it does not explicitly tie each Gemini 2.5 row to a particular Google product or account. These are not Vertex AI quotes. Google’s developer documentation says access to Gemini 2.5 models is limited to people who have actively used them before; for new projects, it points developers to newer models. (Google AI for Developers)
| Gemini 2.5 model | Reported input price per 1M tokens | Reported output price per 1M tokens | Pricing note |
|---|---|---|---|
| Gemini 2.5 Pro | $1.25 up to 200K prompt tokens; $2.50 above 200K | $10 up to 200K; $15 above 200K | Eesel-reported Gemini API rates; its source does not establish how the threshold applies to every Google product or request. |
| Gemini 2.5 Flash | $0.30 for text, image, or video input, as reported by eesel | $2.50 | Eesel-reported Gemini API rates. The source does not specify separate input rates by modality. |
| Gemini 2.5 Flash-Lite | $0.10 for text, image, or video input, as reported by eesel | $0.40 | Eesel-reported rates. Separately, Requesty’s Gemini API listing reports the same provider rates; its listing does not settle any applicable intermediary fee. |
All amounts are USD per 1 million tokens; input and output are priced separately. The Pro and Flash rates are reported in eesel’s Gemini API table, and the Flash-Lite row is also listed by Requesty for its Gemini API route. These sources do not verify current official Google rates or establish that the same schedule applies across Google products. Eesel reports Pro prompt-length tiers, but the available source does not explain precisely how Google applies the 200K threshold. Check Google’s pricing page for applicable current terms; that page is not the source for the third-party rates above.
Important: model access depends on the Google channel
Google’s Gemini Developer API model pages say 2.5 access is being limited to users who have actively used those models, while stating they remain available through that API until further notice. Google’s Cloud documentation separately says Gemini 2.5 Flash on Vertex AI was scheduled for retirement on October 20, 2026. Those statements concern different channels; don’t assume the Vertex AI notice also means the Developer API has shut down, or that Developer API availability means Vertex AI access remains open. For new production work, Google points developers toward newer models. (Google Cloud documentation)
How to estimate Gemini 2.5 API spend
Use this calculation for a request or a group of requests:
Estimated token charge = (input tokens ÷ 1,000,000 × input rate) + (output tokens ÷ 1,000,000 × output rate)
For an illustrative 1 million text-only requests, each using 1,000 text input tokens and 500 output tokens, the arithmetic using the reported rates is:
| Model and rate assumption | Input charge | Output charge | Estimated total |
|---|---|---|---|
| Gemini 2.5 Flash-Lite: $0.10 input / $0.40 output per 1M | $100 | $200 | $300 |
| Gemini 2.5 Flash: $0.30 input / $2.50 output per 1M | $300 | $1,250 | $1,550 |
| Gemini 2.5 Pro: $1.25 input / $10 output per 1M, assuming the up-to-200K tier applies | $1,250 | $5,000 | $6,250 |
These are hypothetical text-token-only estimates, not Google quotes or invoice predictions. They assume the same text-token volume on every request, use the third-party-reported prices above, and exclude any applicable taxes, intermediary fees, caching, batch treatment, grounding, or other charges. They do not estimate image, audio, or video usage. The Pro calculation assumes its lower reported price tier applies; verify the applicable threshold and channel before budgeting.
For your own estimate, replace the example’s request count and average input/output tokens with measured workload data. Count all billed input—including repeated instructions and context—rather than only the user’s latest message. Then check the applicable Google pricing terms and any intermediary’s fees separately.
What else can change the bill?
- Access route and eligibility: Google’s current Developer API pages restrict Gemini 2.5 access to prior active users. A third-party listing can report a rate without proving that your project can use the model or that the rate applies to your Google account or channel. (Google AI for Developers)
- Batch processing: Google’s Developer API pricing page describes the paid tier as including Batch API with a 50% cost reduction. That general statement does not, by itself, confirm Gemini 2.5 eligibility or the exact batch rates for a particular model. Confirm both before applying a discount to a budget. (Google AI Studio pricing)
- Caching and other features: Third-party sources list caching, grounding, and text-to-speech charges, but the captured evidence does not establish a complete, current official Gemini 2.5 schedule for those features. Don’t add or subtract those amounts from the examples without confirming the model, API channel, eligibility, and billing unit.
- Intermediary fees: If you call through a router or platform, distinguish its fee from Google’s upstream token rate. Fee terms can vary by plan, and third-party pages may not agree about markup.
Which model should you budget for?
On the reported rates alone, Flash-Lite has the lowest token cost, Flash costs more, and Pro costs the most—with a higher reported Pro tier for longer prompts. Price alone does not establish which model will meet your quality, latency, or reliability needs. Google describes Flash-Lite as suited to high-volume, lightweight tasks such as classification and simple extraction, but its documentation recommends newer models for new projects. If you are starting a production integration now, compare an available newer model rather than choosing Gemini 2.5 on price alone. (Google AI for Developers)
Bottom line: Use the figures in the table as planning references, not confirmed current Google rates. Gemini 2.5 access is restricted for new users on the Developer API, and Vertex AI has a separate retirement notice for Flash. Verify the exact model, Google channel, availability, rate schedule, and any platform fee before committing a production budget.
Sources
Google Gemini pricing article, including reported Gemini 2.5 API rates — eesel AI
Gemini 2.5 Flash-Lite API pricing — Requesty
Gemini Developer API pricing — Google AI Studio