Gemini API Pricing (2026)
Google Gemini API pricing for 2026: per-million-token rates for 2.5 Pro, Flash, and Flash-Lite, context caching, and the batch discount.
Google Gemini is the price leader at the low end in 2026: Flash-Lite at $0.10 per million input tokens, Flash at $0.30, and Pro at $1.25 for prompts under 200k tokens. Context caching and a 50% batch discount cut it further. Use Flash-Lite for high-volume work, Flash for everyday production, and Pro for long-context reasoning. Rates below are updated 2026 from Google’s pricing page.
Per-million-token rates (updated 2026)
| Model | Input | Output | Use it when |
|---|---|---|---|
| Gemini 2.5 Pro (≤200k) | $1.25 | $10.00 | Long-context reasoning at a good price |
| Gemini 2.5 Pro (>200k) | $2.50 | $15.00 | Very large prompts |
| Gemini 2.5 Flash | $0.30 | $2.50 | Best everyday price-to-quality |
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | Cheapest capable option for high volume |
Audio input is priced higher than text on Flash and Flash-Lite. Gemini 2.0 Flash was deprecated and shut down on June 1, 2026, so ignore older guides that still quote it.
Caching and batching
- Context caching is cheap: a few cents per million tokens plus an hourly storage fee. It is well suited to resending the same large document.
- Batch tier gives a 50% reduction on input and output for asynchronous jobs.
Honest read
On raw input price, Gemini Flash-Lite is the model to beat in 2026, and Flash hits an excellent quality-per-dollar point for production traffic. The trade-offs are ecosystem and the periodic model churn (the 2.0 Flash shutdown is a recent example), so build so you can swap models. If you want a long context window without frontier pricing, Pro is often cheaper than the comparable OpenAI and Anthropic tiers.
Related
See the cheapest LLM API comparison for the cross-provider table and cost control for the techniques. Background: tokens and context windows. For integration help, see AI integration.