PROVIDER GUIDE · VERIFIED 13 JUL 2026

Google Gemini API pricing, explained

Gemini is the value play at every tier we track — but its real advantage shows up in context window size, not just the per-token rate.

Model Input /1M Output /1M Context
Gemini 2.5 Flash$0.15$0.601M
Gemini 3 Flash$0.50$3.001M
Gemini 3.1 Pro$2.00$12.002M

The context window is the real story

Gemini 2.5 Flash and Gemini 3 Flash both ship with a 1M-token context window at budget and mid-tier prices — matching or beating the context window of models costing several times more from other providers. Gemini 3.1 Pro goes further still: a 2M-token window, the largest of any model we track, at a flagship price that undercuts both Claude Opus 4.8 and GPT-5.5. If your task genuinely needs to reason over a large document, codebase, or conversation history, Gemini's context headroom is often the deciding factor before price even enters the conversation.

Gemini 2.5 Flash vs 3 Flash — the upgrade is real, so is the cost jump

Moving from Gemini 2.5 Flash to Gemini 3 Flash is roughly a 3x input and 5x output price increase for what Google positions as a meaningful capability step up. If 2.5 Flash is already handling your task reliably, there's no need to pay for the upgrade. If you're seeing accuracy issues on 2.5 Flash before reaching for a different provider's mid-tier model, testing 3 Flash first is usually the cheaper experiment.

Gemini 3.1 Pro as the value flagship

At $2.00/$12.00, Gemini 3.1 Pro is priced below both Claude Opus 4.8 ($5.00/$25.00) and GPT-5.5 ($5.00/$30.00) while offering the largest context window of the three. The trade-off is that "flagship" positioning and actual task performance don't always move in lockstep across providers — worth running your own head-to-head on a representative task sample rather than choosing on price and context size alone.

How we'd actually pick

Prices verified against ai.google.dev/gemini-api/docs/pricing, 13 July 2026. See full methodology. Use the calculator to compare Gemini against every other tracked provider for your own usage.