Gemini Pricing in 2026: Plans, API Costs and the Coming Price Rise
By Adam, AI Engineer · 1 November 2026 · 4 min read
Gemini is free to start, Google AI Pro costs $19.99 a month in the US, and AI Ultra starts at $99.99. The Gemini API is billed per token, from $0.10 per million input tokens on 2.5 Flash-Lite to $2 on 4 Argon. Two popular models are about to double in price: Gemini 4 Argon on 1 December 2026 and Gemini 3.8 Flash on 1 January 2027.
Google AI plans for people
These come from Google's subscriptions page, checked on 1 November 2026. Prices vary by country, so check the page for yours.
| Plan | US price per month | Price shown in Pakistan | Storage |
|---|---|---|---|
| Free | $0 | Rs 0 | 15 GB |
| AI Plus | Check your country | Rs 1,400 | 400 GB |
| AI Pro | $19.99 | Rs 5,600 | 5 TB |
| AI Ultra | From $99.99; top tier $199.99 | From Rs 21,999; top tier Rs 55,900 | From 20 TB |
AI Pro gives four times the usage limits of Free. The two Ultra tiers give 5x and 20x the limits of AI Pro. The storage matters too: if you already pay for Google storage, the true extra cost of AI Pro may be lower than it looks.
Gemini for teams
Google lists Gemini Enterprise Business at about $21 per seat a month, for up to 300 seats. Standard and Plus are listed at about $30 per seat a month, with higher usage limits.
Gemini API prices
These are paid-tier prices per million tokens from the Gemini API pricing page. The last column uses one example: 100,000 requests a month, each with 3,000 input tokens and 400 output tokens, and no caching.
| Model | Input | Output | Example month | Change |
|---|---|---|---|---|
| 2.5 Flash-Lite | $0.10 | $0.40 | $46 | None listed |
| 3.1 Flash-Lite | $0.25 | $1.50 | $135 | None listed |
| 3.5 Flash-Lite | $0.30 | $2.50 | $190 | None listed |
| 3.8 Flash | $0.75 | $3.75 | $375 | $750 from 1 Jan 2027 |
| 4 Argon | $2 | $10 | $1,000 | $2,000 from 1 Dec 2026 |
What the price rise means
From 1 December, Gemini 4 Argon goes to $4 input and $20 output. From 1 January, Gemini 3.8 Flash goes to $1.50 input and $7.50 output. If you use either in a product, recalculate your margin now and test an alternative before the change. Our AI price update explains what to check.
How to pay less
- Free tier: 3.8 Flash and the Flash-Lite models above have a free API tier, which is useful for testing. 4 Argon does not.
- Batch processing: Google lists a 50% reduction for delayed jobs, so the 3.8 Flash example falls to about $188 at today's price.
- Choose Flash-Lite for simple work: tagging and short summaries may not need a bigger model. Run the same test set on each and compare quality.
Costs that are easy to miss
- Google Search grounding: on Gemini 3 and newer models, the paid tier includes 5,000 free requests a month, then charges $14 per 1,000.
- Cache storage: on 3.8 Flash, cached content has a storage fee of $0.50 per million tokens per hour through 31 December 2026, then $1.00.
- Thinking tokens: Google's output prices include thinking, so a model that thinks a lot can use more output than the visible answer.
Which option should you choose?
- You just use it yourself: start with Free. Choose AI Pro if you also want the storage or keep hitting limits.
- You are building a product: test on the free API tier, then budget using next year's prices rather than today's.
Compare Gemini with GPT and Claude on your own numbers in the free AI cost calculator, or read our Claude pricing and GPT pricing guides.