Gemini 3 Flash Preview API pricing
Gemini 3 Flash Preview costs $0.50 per million input tokens and $3 per million output tokens at the standard rate. Here are its cached and batch prices, and what it costs for your usage.
| Token type | Standard | Batch (50% off) |
|---|---|---|
| Input | $0.50 | $0.25 |
| Cached input | $0.05 | $0.05 |
| Output | $3 | $1.50 |
Gemini 3 Flash Preview: Audio input costs more.
- Input $15.00
- Cached input $1.50
- Output $27.00
| Workload | Standard | Batch | |
|---|---|---|---|
Support chatbot 2,000 in / 300 out tokens,
1,000 requests a day, 50% cached | $43.50 | $22.50 | 6th cheapest of 18 |
Document summaries 10,000 in / 800 out tokens,
200 requests a day | $44.40 | $22.20 | 6th cheapest of 18 |
Tagging / classifying 400 in / 20 out tokens,
20,000 requests a day | $156.00 | $78.00 | 6th cheapest of 18 |
Compare Gemini 3 Flash Preview with
Other Google models
FAQ
Frequently asked questions
How much does Gemini 3 Flash Preview cost per million tokens?
Gemini 3 Flash Preview costs $0.50 per million input tokens and $3 per million output tokens at the standard rate. Prices are in USD, before tax, from Gemini API pricing, last checked on 30 Sept 2026. Audio input costs more.
Does Gemini 3 Flash Preview have a Batch API discount?
Yes. Through the Batch API, Gemini 3 Flash Preview costs $0.25 per million input tokens and $1.50 per million output tokens, 50% less than the standard rate. Batch requests are processed asynchronously rather than straight away.
How much does prompt caching save on Gemini 3 Flash Preview?
Input read from the prompt cache costs $0.05 per million tokens instead of $0.50, 90% less. Writing to or storing the cache can cost extra; those charges aren’t included here.
Is Gemini 3 Flash Preview one of the cheaper API models?
For a support chatbot sending 1,000 requests a day (2,000 input and 300 output tokens each, 50% of the input cached), Gemini 3 Flash Preview is the 6th cheapest of the 18 OpenAI, Anthropic and Google models we track, at $43.50 a month. Price alone doesn’t tell you how well a model will handle your task, so test the candidates on your own prompts.