Gemini 3.1 Pro (preview) pricing
Google's newest flagship preview. Dual-tier pricing doubles above 200K input tokens - watch your context size.
Input
$2.00/ 1M tok
Output
$12.00/ 1M tok
- Context window
- 1M
- Max output
- -
- Cached input
- $0.200 / 1M
- Verified
- 2026-04-06
Long-context tier
Above 200K input tokens, this model bills at $4.00 input / $18.00 output per 1M tokens.
Gemini 3.1 Pro (preview) is Google's in-development flagship. Pricing follows the familiar Gemini pattern: a standard rate of $2 per 1M input and $12 per 1M output, then a doubled rate ($4 / $18) once your input crosses 200,000 tokens. That threshold catches a lot of long-document and large-context workflows by surprise, so it's worth knowing before you deploy.
Cached input lands at $0.20 per 1M tokens - a 90% discount on the standard rate. Google's context caching is explicit (you create and reuse cache objects via the API) rather than automatic, so you need to wire it up deliberately, but the savings are real on long repeated prefixes.
Calcis uses Google's own countTokens endpoint so the input figure matches exactly what Google bills against, and applies the long-context surcharge in the estimate whenever your prompt crosses the 200K threshold.
Estimate your cost on Gemini 3.1 Pro (preview)
Paste your prompt into the estimator, pick Gemini 3.1 Pro (preview), and see the exact dollar cost - input tokens counted with the provider's own tokenizer, output tokens predicted by our regression model.
Frequently asked
- When does Gemini 3.1 Pro switch to long-context pricing?
- At 200,000 input tokens. Below that, $2 input / $12 output per 1M. Above, $4 input / $18 output per 1M. The switch applies to the whole request - a 201K-token prompt is billed at the long-context rate, not just the overflow.
- How much does Gemini 3.1 Pro cost per request?
- A 1,000-token prompt with a 500-token reply costs about $0.008 ($0.002 input + $0.006 output). Crossing 200K input doubles both sides - a 300K prompt with a 1K reply costs about $1.22.
- Is 3.1 Pro cheaper than 2.5 Pro?
- Input is more expensive on 3.1 Pro ($2 vs $1.25) but output is cheaper ($12 vs $10 at the standard tier, moving to $18 vs $15 at long-context). Headline costs are close; the differentiator is usually capability.
- What tokenizer does Gemini use?
- Google uses SentencePiece under the hood but doesn't publish it for direct use. Calcis calls Google's countTokens API for the authoritative count - the same boundary Google bills against.
Pricing verified 2026-04-06 from the provider's rate card.