Google positions Gemini API pricing across multiple layers: access tiers such as Free, Paid, and Enterprise, model-specific token rates, and billing-account controls that determine how much a team can spend. For B2B buyers, the practical task is not just comparing per-token prices, but understanding which usage mode, billing tier, and add-on charges apply before a workload moves into production.
What it is and who it’s for
Gemini API is Google’s API for accessing Gemini models through Google AI for Developers tooling, including Google AI Studio. According to Google AI for Developers, access is grouped into Free, Paid, and Enterprise tiers.
The Free tier is aimed at early testing, with limited access to certain models and free input and output tokens. Paid access adds higher rate limits, context caching, Batch API support, and access to advanced models. Enterprise is positioned for organizations that need dedicated support, advanced security and compliance features, provisioned throughput, and volume discounts.
That makes the offering relevant to two broad groups. First are development teams experimenting in Google AI Studio and moving toward live application usage. Second are enterprise teams that expect production traffic, need support commitments, or want pricing terms beyond published paid-tier rates. Teams comparing costs across model providers can also see our Claude API Pricing Guide: Tiers, Caching, and Cost Control.
How it works
Gemini API pricing is metered by usage, but Google breaks that usage into more than one billable component. According to Google AI for Developers billing documentation, Gemini API pricing is based on input tokens, output tokens, cached tokens, and cached storage duration.
A token is the unit Google uses to measure text and other model inputs and outputs for billing. In practice, that means a team’s bill depends on how much prompt data it sends to the model, how much generated output it receives back, and whether it uses features such as caching or grounding.
Google also separates commercial access from billing-account progression. New accounts start on the Free Tier. In billing, Google AI for Developers lists Free, Tier 1, Tier 2, and Tier 3 account levels. Tier 1 requires an active billing account and has a $250 cap. Tier 2 requires $100 paid plus 3 days since the first successful payment and has a $2,000 cap. Tier 3 requires $1,000 paid plus 30 days since the first successful payment and has a $20,000-$100,000+ cap.
Paid usage starts with prepay mechanics. Upgrading to Paid Tier requires linking a billing account and prepaying at least $10. Google says paid-tier requests are served only while the prepay credit balance remains positive, with a minimum credit purchase of $10 and a maximum credit purchase of $5,000.
Teams can also automate replenishment. Google says auto-reload can top up credits when the balance is low, and a monthly auto-charge limit can cap those automatic reloads within a billing cycle. Postpay, where charges are collected at month end or when the spend cap is reached, becomes available once Tier 3 criteria are met.
Pricing and cost considerations
The clearest way to read Gemini API pricing is to split it into three cost layers: model token rates, discounted execution modes such as Batch or Flex, and extras such as caching or grounding.
| Model / mode | Input price | Output price | Notes |
|---|---|---|---|
| Gemini 3.6 Flash Standard | $1.50 per 1M tokens | $7.50 per 1M tokens | Paid tier |
| Gemini 3.6 Flash Batch and Flex | $0.75 per 1M tokens | $3.75 per 1M tokens | Batch API is priced lower than standard |
| Gemini 3.6 Flash Priority | $2.70 per 1M tokens | $13.50 per 1M tokens | Paid tier |
| Gemini 3.5 Flash Standard | $1.50 per 1M tokens | $9.00 per 1M tokens | Paid tier |
| Gemini 3.5 Flash Batch and Flex | $0.75 per 1M tokens | $4.50 per 1M tokens | Paid tier |
| Gemini 3.5 Flash Priority | $2.70 per 1M tokens | $16.20 per 1M tokens | Paid tier |
| Gemini 3.5 Flash-Lite Standard | $0.30 per 1M tokens | $2.50 per 1M tokens | Paid tier |
| Gemini 3.5 Flash-Lite Batch and Flex | $0.15 per 1M tokens | $1.25 per 1M tokens | Paid tier |
| Gemini 3.5 Flash-Lite Priority | $0.54 per 1M tokens | $4.50 per 1M tokens | Paid tier |
| Gemini 3.1 Flash-Lite Standard | $0.25 per 1M text/image/video tokens; $0.50 for audio | $1.50 per 1M tokens | Standard mode |
| Gemini Omni Flash Preview | $1.50 per 1M tokens | $9.00 per 1M tokens (text) or $17.50 per 1M tokens (video) | Paid tier only; video billed at ~$0.10/second of 720p output |
| Gemini 3.1 Pro Preview Standard | $2.00 or $4.00 per 1M tokens | $12.00 or $18.00 per 1M tokens | Price changes above or below 200k tokens |
| Gemini 3.1 Pro Preview Batch/Flex | $1.00 or $2.00 per 1M tokens | $6.00 or $9.00 per 1M tokens | Price changes above or below 200k tokens |
| Gemini 3.1 Pro Preview Priority | $3.60 or $7.20 per 1M tokens | $21.60 or $32.40 per 1M tokens | Price changes above or below 200k tokens |
| Gemini 3.5 Live Translate | $3.50 per 1M input tokens or $0.0053/min audio | $21.00 per 1M output tokens or $0.0315/min audio | Live translation pricing |
Two pricing mechanics matter for cost control. First, Google states that Batch API pricing is 50% lower than standard pricing. That can materially change the cost of non-urgent workloads such as offline processing or queued generation jobs.
Second, some services add separate charges outside the base model call. Context caching adds a per-token caching rate on top of the base input price, plus separate storage pricing per 1,000,000 tokens per hour. Teams that want a broader frame for consumption controls across providers can also see our OpenAI Pricing Guide: ChatGPT Plans and API Usage Controls. Grounding with Google Search and Google Maps includes 5,000 free operations per month, shared across Gemini 3 models, before billing at $14 per 1,000 search queries; Google labels that free quota “prompts” for Standard-mode pricing tables and “requests” for Batch, Flex, and Priority tables, but the $14-per-1,000 rate and per-individual-query billing apply the same way across modes.
How to choose
The first filter is workload shape. If a team is evaluating prototypes or internal tests, the Free tier exists for that stage, but Google describes access as limited and does not enumerate every free-tier model option on the pricing page.
The second filter is latency and scheduling tolerance. Paid users can choose between Standard, Batch and Flex, and Priority options on several models. Standard is the baseline paid path, Batch and Flex are the lower-cost option where turnaround can be deferred, and Priority is the higher-cost option for traffic that needs that service class.
The third filter is model class. Teams focused on lower-cost usage may gravitate toward Flash-Lite variants, while teams using larger prompts need to account for models such as Gemini 3.1 Pro Preview that change price above or below the 200k-token threshold.
The fourth filter is buying model. Buyers standardizing spend controls across multiple SaaS vendors can also see our SaaS Pricing Models: How Buyers Should Evaluate Cost. Enterprise access is the route Google presents for organizations that need provisioned throughput, dedicated support, advanced security and compliance, and volume discounts.
Limitations and gotchas
Gemini API pricing has several practical constraints that matter during rollout.
Billing access tiers and account spend tiers are related but not identical. Google describes product access as Free, Paid, and Enterprise, while billing is managed through Free, Tier 1, Tier 2, and Tier 3 account levels with different caps across all linked projects.
Those caps apply at the billing-account level, not just to a single application. Google says billing account tier caps apply across all linked projects, and service pauses for all linked projects when the billing account tier cap is reached.
Credit handling also matters. Prepay credits expire after 12 months and are non-refundable except after switching to Postpay. Prepay is not available for Invoiced or Offline accounts.
Teams should also avoid assuming Google Cloud trial money offsets Gemini usage. Google says the Google Cloud Welcome credit is $300, but it cannot be used for Gemini API or AI Studio.
Operational visibility is not instant. Google says billing signals can lag, with tier upgrades usually reflecting within 10 minutes, and cost breakdown graphs can take up to 24 hours to update.
FAQ
Does using Batch or Priority mode change grounding costs?
No. The $14-per-1,000-query rate and the free monthly allowance apply the same way across Standard, Batch, Flex, and Priority modes. Google’s pricing tables just label that shared free quota “prompts” in Standard-mode tables and “requests” in Batch, Flex, and Priority tables.
Can Enterprise pricing differ from the published per-token rates?
Yes. Google notes that prices offered through the Gemini Enterprise Agent Platform may differ from the rates listed on the standard Gemini API pricing page, so enterprise buyers should confirm terms directly rather than assuming published token rates apply.
What happens to auto-reload and unused credits when an account switches to Postpay?
Google says upgrading to Postpay closes the Prepay payments account, turns off auto-reload, and automatically refunds any unused Prepay credits, subject to standard refund processing.
Does Google AI Studio itself cost anything separate from API usage?
No. Google says Google AI Studio usage is free of charge in all available regions; charges apply to the underlying Gemini API calls, not to using the Studio interface.
