Every current Gemini model with its real per-million-token price — plus the free tier, the long-context surcharge, a calculator for your own workload, and a straight comparison against GPT and Claude.
| MODEL | INPUT $/1M | CACHED IN | OUTPUT $/1M | BLENDED | CONTEXT | YOUR COST |
|---|---|---|---|---|---|---|
Gemini 3.1 Pro gemini-3.1-pro-preview · >200K: $4/$18 | $2.00 | $0.20 | $12.00 | $4.50 | 1.05M | $8.00 |
Gemini 3.8 Flash gemini-3.8-flash · promo to Dec 31 | $0.75 | $0.075 | $3.75 | $1.50 | 1.05M | $2.63 |
Gemini 3.7 Flash gemini-3.7-flash · promo to Dec 31 | $0.75 | $0.075 | $3.75 | $1.50 | 1.05M | $2.63 |
Gemini 3.5 Flash gemini-3.5-flash · prior gen | $1.50 | $0.15 | $9.00 | $3.38 | 1.05M | $6.00 |
Gemini 3.5 Flash-Lite gemini-3.5-flash-lite · cheapest tier | $0.30 | $0.03 | $2.50 | $0.85 | 1.05M | $1.55 |
USD per 1M tokens from Google’s official pricing page as of October 16, 2026. Blended assumes a 3:1 input:output ratio. Get keys at the Google console.
Gemini is the only one of the big three with a genuinely free API tier: AI Studio keys run at no cost within rate limits, which is why so many prototypes start here. Paid rates are aggressive too — Gemini 3.8 Flash at $0.75 input / $3.75 output per 1M tokens is among the cheapest current-generation multimodal workhorses on the market, and every current Gemini model takes text, images, audio, and video in a 1M-token context. Budget ahead, though: that Flash rate is promotional through December 31, 2026, after which 3.6, 3.7, and 3.8 Flash double to $1.50 / $7.50.
The catch to budget for: long-context surcharges. Prompts beyond 200K tokens bill at a higher rate on Gemini 3.1 Pro — $4 input / $18 output instead of $2 / $12. If you routinely stuff half the context window, your effective rate is closer to double the sticker price.
Context caching bills cache reads at about a tenth of the fresh input rate, plus an hourly storage fee for explicit caches that varies by model — $0.50 per million tokens per hour on the current Flash models (rising to $1.00 in January), $1.00 on 3.5 Flash and Flash-Lite, and $4.50 on 3.1 Pro. Implicit caching kicks in automatically on prompts over ~4K tokens. The Batch API halves prices for asynchronous jobs — worth it for bulk multimodal processing, where Gemini is usually the cheapest option anyway.
Gemini bills per token with separate input and output rates per million tokens: Gemini 3.1 Pro at $2 / $12 (rising to $4 / $18 beyond 200K prompt tokens), 3.8 Flash at $0.75 / $3.75 (promotional through December 31, 2026), and 3.5 Flash-Lite at $0.30 / $2.50. All current models handle text, images, audio, and video.
Yes, within limits: AI Studio API keys have a free tier on most models (3.1 Pro is paid-only) with rate limits that are enough for prototypes and small tools. Free-tier prompts may be used to improve Google’s products, so production traffic belongs on the paid tier, billed per token at the rates in the table above.
On Gemini 3.1 Pro, any request whose prompt exceeds 200K tokens bills the whole request at the higher tier — $4 input / $18 output instead of $2 / $12. Flash models keep flat pricing. If you regularly send huge contexts, price your workload at the surcharge rate.
Mostly, tier for tier: Gemini 3.1 Pro ($4.50/1M blended) undercuts Claude Opus 5.5 ($8) and sits just above GPT-6 Sol ($4), while 3.8 Flash ($1.50) undercuts both mid-tier rivals by more than half — as long as you stay under the 200K long-context threshold on Pro. The matchup table below shows all three side by side.
Start with 3.8 Flash — the newest generation, fully multimodal, and a fraction of Pro pricing. Move up to 3.1 Pro when answer quality on hard reasoning is worth 3× the cost, or down to Flash-Lite for bulk classification at $0.30 / $2.50.
Also see: OpenAI API pricing · Claude API pricing · Grok API pricing · DeepSeek API pricing · Kimi API pricing · GLM API pricing · MiniMax API pricing · MiMo API pricing · all 68 models
ReqKey gives every key a credit balance — validate, deduct, enforce limits, and watch per-consumer Gemini spend in real time.