Gemini API Pricing (2026)
Google prices the Gemini API per million tokens, with one wrinkle the other providers do not have: Gemini's Pro models charge different rates depending on how large your prompt is. This page lists the current rates and explains that tier split.
Gemini model prices
List prices verified 2026-10-11. USD per 1 million tokens, standard tier.
| Model | Prompt size | Input $/1M | Cached input $/1M | Output $/1M | Context |
|---|---|---|---|---|---|
| Gemini 3.8 Flash (promo price) | Any | $0.75 | $0.075 | $3.75 | 1,048,576 |
| Gemini 3.6 Flash (promo price) | Any | $0.75 | $0.075 | $3.75 | 1,048,576 |
| Gemini 3.5 Flash-Lite | Any | $0.30 | $0.03 | $2.50 | 1,048,576 |
| Gemini 3.1 Pro (preview) | Up to 200K tokens | $2.00 | $0.20 | $12.00 | 1,048,576 |
| Gemini 3.1 Pro (preview) | Above 200K tokens | $4.00 | $0.40 | $18.00 | 1,048,576 |
| Gemini 3.1 Flash-Lite | Any | $0.25 | $0.025 | $1.50 | 1,048,576 |
| Gemini 2.5 Pro | Up to 200K tokens | $1.25 | $0.125 | $10.00 | 1,048,576 |
| Gemini 2.5 Pro | Above 200K tokens | $2.50 | $0.25 | $15.00 | 1,048,576 |
| Gemini 2.5 Flash | Any | $0.30 | $0.03 | $2.50 | 1,048,576 |
| Gemini 2.5 Flash-Lite | Any | $0.10 | $0.01 | $0.40 | 1,048,576 |
Every model in the table takes 1,048,576 tokens of context with up to 65,536 tokens of output. Gemini 3.8 Flash and 3.6 Flash are on promotional pricing through 2026-12-31; from 2027-01-01 Google lists $1.50 input, $0.15 cached input, and $7.50 output. Context caching also charges hourly storage per cached token, which the calculator does not model.
How the 200K price tier works
Gemini 2.5 Pro has two price tiers keyed to prompt size. If your prompt is 200K tokens or less, input bills at $1.25 per 1M and output at $10.00. If the prompt exceeds 200K tokens, the higher tier applies: $2.50 input and $15.00 output.
Note what the trigger is. The tier is set by the size of the prompt, and once a prompt crosses 200K tokens, the higher output rate applies to that request too. Crossing the line doubles the input rate and raises the output rate by half. Gemini 3.1 Pro (preview) uses the same 200K split: $2.00 / $12.00 up to 200K tokens, $4.00 / $18.00 above.
This makes prompt size a budget decision on 2.5 Pro. The model accepts prompts up to 1,048,576 tokens, but a request that stuffs the full context costs twice as much per input token as one that stays under 200K. If a long-document job can be split so each request stays at or below 200K tokens, the same total token volume bills at the lower tier. When a request genuinely needs more than 200K tokens of context at once, the higher tier is the price of that capability.
Counting Gemini tokens
Google exposes a free countTokens API, so you can get an exact count for a prompt before sending it. That matters more on Gemini than elsewhere because of the tier split: an estimate that is wrong by a few thousand tokens near the 200K line is the difference between $1.25 and $2.50 per 1M on the whole request. Count first when a prompt is anywhere near the boundary. The Gemini token counter covers this.
Where Flash fits
Gemini 2.5 Flash prices at $0.30 input and $2.50 output with the same 1,048,576-token context, and it has no tier split. For high-volume work that does not need Pro, the flat rate also removes the need to watch the 200K boundary.
Related pages
- All-provider price comparison, every model in one table.
- Claude API pricing and OpenAI API pricing.
- Gemini 2.5 Pro context window in detail.