Gemini API pricing is charged per 1 million tokens, across a Free tier, a Paid tier and Enterprise, and the paid price depends on the model. As of October 2026, Google's pricing page lists Gemini 4 Argon at $2.00 for input and $10.00 for output per 1M tokens through November 30, 2026, rising to $4.00 and $20.00 from December 1, 2026.
That step-up is the detail most people miss. Several Gemini models carry an introductory price with an end date printed next to it, so a cost you work out today may not hold next quarter.
Free, Paid and Enterprise: how the tiers differ
| Tier | What Google lists | Is your content used to improve Google's products? |
|---|---|---|
| Free | Limited access to certain models, free input and output tokens, Google AI Studio access | Yes |
| Paid | Higher rate limits, context caching, Batch API at half the cost, Google's most advanced models | No |
| Enterprise | Everything in Paid, plus optional dedicated support, advanced security and compliance, provisioned throughput and volume-based discounts | No, as with Paid |
Moving to Paid means linking a billing account in Google AI Studio and, on the Prepay plan, adding a minimum of $5 (or the equivalent in other currencies) of credits. Your usage is then deducted from that balance in near real time.
Two Prepay rules are worth knowing before you top up. Unused credits expire after 12 months and are non-refundable, except after switching to a Postpay account. And when your Prepay balance hits zero, every API key in every project linked to that billing account stops working at the same time, with requests failing as HTTP 402 errors. If a client-facing tool runs on the API, that's the line to watch.
What a million tokens costs on three models
These are Google's standard paid-tier prices per 1M tokens, shown in dollars on its pricing page as of October 2026:
| Model | Free tier | Paid input | Paid output (including thinking tokens) |
|---|---|---|---|
| Gemini 4 Argon | Not available | $2.00 through November 30, 2026, then $4.00 | $10.00 through November 30, 2026, then $20.00 |
| Gemini 3.8 Flash | Free of charge | $0.75 through December 31, 2026, then $1.50 | $3.75 through December 31, 2026, then $7.50 |
| Gemini 3.5 Flash-Lite | Free of charge | $0.30 | $2.50 |
Google describes Argon as its flagship reasoning model for complex software engineering, deep problem solving and autonomous multi-step agents. Flash-Lite is the cost-efficient one, aimed at high-volume tasks, translation and simple data processing. For most business automations, like sorting enquiries or summarising documents, the gap between those two prices is what you're deciding about.
Two dates to put in your diary
- December 1, 2026: Gemini 4 Argon's input and output prices double on the standard paid tier.
- January 1, 2027: Gemini 3.8 Flash's input and output prices double too, to $1.50 and $7.50.
- No date listed: Gemini 3.5 Flash-Lite's paid prices carry no step-up on Google's page as of October 2026.
If you're quoting a client for an AI-powered workflow, build those dates into the quote. I won't guess your bill for you here, because it depends entirely on how many tokens your prompts and outputs use.
Thinking tokens and grounding: the extra lines on the bill
Google's output prices include thinking tokens, the reasoning a model does before it answers. For Argon, the pricing page goes further: every token the model processes is counted and billed, including thinking tokens and internal tool system instructions and function declarations when you use grounding.
Grounding with Google Search has its own meter. On Gemini 3 and newer models you get 5,000 free search requests per month, shared across those models, then $14 per 1,000 requests, with a daily cap of 1,500 requests. Google also notes that one request from your customer may result in more than one Google Search query, and each query is charged.
Batch pricing: half price if you can wait
The Batch API runs requests in bulk rather than live, and it's priced at half the standard rate. For Argon that's $1.00 input and $5.00 output per 1M tokens through November 30, 2026, then $2.00 and $10.00. Batch jobs have their own rate limits, including up to 100 concurrent batch requests and a 2GB input file limit. Overnight jobs, like tagging a product catalogue or summarising last month's support tickets, are where batch earns its keep.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Spend limits by usage tier
Google also caps how fast you can spend, to protect you from surprise charges. These limits are checked over a rolling 10-minute window, and your tier rises automatically as your spending history grows:
| Usage tier | How you qualify | Spend limit per 10 minutes | Billing tier cap |
|---|---|---|---|
| Free | Active project or free trial | N/A | N/A |
| Tier 1 | Set up and link an active billing account | $10 | $250 |
| Tier 2 | Paid $100 plus 3 days from first successful payment | $50 | $2,000 |
| Tier 3 | Paid $1,000 plus 30 days from first successful payment | $200 | $20,000 to $100,000+ |
Hit a spend limit and the API returns a 429 RESOURCE_EXHAUSTED error. Google's fixes are to wait and retry, cut back on expensive requests with smaller context windows or shorter outputs, or request a rate limit increase. Rate limits apply per project, and requests-per-day quotas reset at midnight Pacific time.
How I'd budget for it
I'd start on the free tier to prototype, and skip the paid tier until you know your real token volume from a working test. One exception: skip the free tier for anything containing client data, because Google says free-tier content is used to improve its products and paid-tier content isn't. If you're still choosing a provider, start with what the Gemini API's free tier covers and this comparison of free AI API keys. If you only want to chat with Gemini rather than build on it, an app plan such as Google AI Pro is a different product. Prices change, so always check Google's pricing page before you commit.
If you want to forecast what an AI-powered workflow will cost before you build it, that's the kind of AI enablement I help with. I run AI training and workshops for teams, and help businesses with AI enablement and workflow automation, so the busywork runs itself and people get time back. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.
Frequently asked questions
Is the Gemini API free?
There's a Free tier with limited access to certain models and free input and output tokens, but Gemini 4 Argon isn't available on it, and Google says free-tier content is used to improve its products.
How much does Gemini 4 Argon cost per million tokens?
As of October 2026, on the standard paid tier: $2.00 input and $10.00 output through November 30, 2026, then $4.00 and $20.00 from December 1, 2026.
Are thinking tokens billed on the Gemini API?
Yes. Google's output prices include thinking tokens, and for Argon it says every token the model processes is counted and billed.
What does a 429 RESOURCE_EXHAUSTED error mean?
You've hit a rate limit, which can include the spend-based limit for your usage tier. Wait and retry, reduce expensive requests, or request a rate limit increase.
