Requesty
Back|NOV '26AI MODELS / COST OPTIMIZATION
7 MIN READ|

Gemini API pricing: model rates, caching and free tiers

Last updated

In November 2026, Gemini API Standard pricing for new projects starts at $0.25 per million input tokens and $1.50 per million output tokens with Gemini 3.1 Flash-Lite. Gemini 3.8 Flash costs $0.75 and $3.75, while Gemini 4 Argon costs $2 and $10 through November 30. Several rates double in the coming weeks.

Gemini API prices in November 2026

The table below lists Google's Standard rates in US dollars per million tokens. Rates are for text input and text output; audio and other media can be priced differently. Google's pricing documentation says thinking tokens are billed too.

ModelText input per 1M tokensText output per 1M tokensPrice change or context
Gemini 4 Argon$2.00$10.00$4.00 input and $20.00 output from December 1, 2026
Gemini 4 Carbon$2.00$10.00$4.00 input and $20.00 output from December 1, 2026
Gemini 4 Hydrogen$1.50$7.50Flat rate across its 1 million-token context
Gemini 3.8 Flash$0.75$3.75$1.50 input and $7.50 output from January 1, 2027
Gemini 3.6 Flash$0.75$3.75$1.50 input and $7.50 output from January 1, 2027
Gemini 3.5 Flash-Lite$0.30$2.50Standard rate
Gemini 2.5 Flash$0.30$2.50Existing users only; calls served by 3.5 Flash-Lite from December 1, 2026
Gemini 3.1 Flash-Lite$0.25$1.50Standard rate
Gemini 2.5 Flash-Lite$0.10$0.40Standard rate
Gemini 3.1 Pro Preview$2.00$12.00Up to 200,000 tokens; $4.00 input and $18.00 output above that
Gemini 2.5 Pro$1.25$10.00Up to 200,000 tokens; $2.50 input and $15.00 output above that

Google's deprecations page limits Gemini 2.5 models to users who have actively used them and points new projects to 3.5 Flash-Lite or 3.8 Flash. For a new project, the cheapest model in the table is Gemini 3.1 Flash-Lite, at $0.25 input and $1.50 output per million tokens.

Source: Google's Gemini API pricing, November 2026.

Argon has one rate across its context window. That simplifies long-context estimates compared with 2.5 Pro or 3.1 Pro Preview, where crossing 200,000 tokens changes the rates.

The endpoint changes the bill too. Requesty's Vertex US listing for Hydrogen is $1.65 per million input tokens and $8.25 per million output tokens, versus $1.50 and $7.50 for the Gemini API in the table.

Check model changes before December 1

From December 1, 2026, calls to gemini-2.5-pro are scheduled to be served by gemini-3.8-flash. Calls to gemini-2.5-flash are scheduled to be served by gemini-3.5-flash-lite, with each billed at the new model's rates. See Google's deprecations page.

Test the replacement on your own prompts before then. A lower token price alone does not tell you whether the response will fit your application.

Google's changelog also says requests to gemini-3.7-flash now route to gemini-3.8-flash, and the older model can no longer be selected in AI Studio.

What caching saves on Gemini 3.8 Flash

Prompt caching reuses an eligible part of a prompt, so a cached read costs less than sending those tokens at the regular input rate. Gemini offers implicit caching, which is automatic with no storage charge, and explicit caching, where you create a cache and pay hourly storage.

Through December 31, 2026, Gemini 3.8 Flash cached reads cost $0.075 per million tokens, while explicit cache storage costs $0.50 per million tokens per hour. On January 1, 2027, the cached read rate rises to $0.15 and explicit cache storage rises to $1.00.

With implicit caching, a hit gets the cached read rate with no storage cost. For an explicit-cache example, take a 25,000-token prefix stored for one hour. The example uses one cached read during that hour, November 2026 Standard rates, and no costs beyond the read and storage.

At $0.75 per million tokens, a regular input read costs 25,000 / 1,000,000 x $0.75 = $0.01875. The cached read costs 25,000 / 1,000,000 x $0.075 = $0.001875, and one hour of explicit cache storage costs 25,000 / 1,000,000 x $0.50 = $0.0125.

The cached read saves $0.016875, leaving about $0.0044 after that hour of storage. A second read in the same hour would add another $0.016875 in savings. This example leaves out the rest of the request and the cost of creating the cache.

Implicit caching has a minimum of 4,096 tokens for Argon and 3.8 Flash. A cache hit is not guaranteed, as covered in our cache read price comparison.

Argon's implicit cached input rate is $0.10 per million tokens through November 30, then $0.20 from December 1. Argon does not support explicit caching.

What Gemini costs in routed traffic

List prices help with estimating a new workload. Once an application is running, its input and output mix, request length and cache use will affect the bill.

Our published rankings cover routed traffic across Requesty. Blended cost per million tokens is what Requesty billed divided by the total input and output tokens served, expressed per million tokens, with prompt cache discounts applied. Cache hit rate is cached input tokens divided by input tokens, as defined in our provider data note.

ModelBlended cost per 1M tokensCache hit rate
Gemini 3.8 Flash$0.4753.6%
Gemini 4 Argon$0.8770.9%
Gemini 3.5 Flash-Lite$0.3318.2%
Gemini 3.1 Flash-Lite$0.359.1%

The requests behind these figures are different workloads, so request size and cache use vary across the model rows.

Our provider cache hit rate note covers October 1 to October 31, 2026. Vertex (Gemini) had a 51.06% cache hit rate, compared with 77.73% for OpenAI and 94.04% for Fireworks. These provider figures cover different workloads, and providers can report cached tokens differently; the model figures in the table combine traffic across providers.

Per-token price can also diverge from cost per task: a model with a lower token price can use more tokens to finish the same job. Run the same task on the models you are considering and record total token use, including thinking tokens.

Batch, Flex and Priority change the rate

Batch is half the Standard price for work that can wait up to a day. Flex uses Batch pricing for on-demand calls that may be shed. Priority costs about 1.8 times Standard.

For Argon, Google lists the following rates per million tokens:

TierInput through November 30, 2026Output through November 30, 2026Input from December 1, 2026Output from December 1, 2026
Standard$2.00$10.00$4.00$20.00
Batch and Flex$1.00$5.00$2.00$10.00
Priority$3.60$18.00$7.20$36.00

The tier also explains a common 3.8 Flash quote. In November 2026, Standard input costs $0.75 per million tokens, while the $1.35 input rate applies to Priority. Choose the tier before comparing price cards.

Which Gemini models have a free tier?

Gemini 4 Hydrogen has a free tier, as do Gemini 3.8 Flash, 3.6 Flash and 3.5 Flash-Lite, among other Flash models. Argon, Carbon and 3.1 Pro Preview do not.

Google's rate limits apply per project, and daily request quotas reset at midnight Pacific time. Check your model's limits in Google AI Studio before sizing a workload. For a wider comparison, see our guide to LLM API rate limits by provider.

Data terms also differ. Google's pricing page says free-tier content is used to improve its products, while paid-tier content is not. Check Google's terms for your region.

Grounding with Google Search is available on the paid tier. Gemini 3 and newer models share an allowance of 5,000 free grounding requests per month, followed by a charge of $14 per 1,000 requests. The default cap is 1,500 requests per day.

Estimate your own bill

Start with a sample of requests from the application you plan to run. Record input tokens, output tokens and cached tokens separately, then apply the rate for the model, endpoint and service tier you intend to use. To compare Gemini with other providers, use our LLM API price guide.

Before you set a Gemini budget

0 of 4 done

Requesty
Compare models through one API

Requesty is an LLM gateway with one OpenAI-compatible API for more than 600 models, including Gemini 3.8 Flash. You can route requests across providers and use cost analytics to compare the resulting spend. Pay as you go adds 5% to model costs.

Frequently asked questions
How much does the Gemini API cost?
In November 2026, Gemini 2.5 Flash-Lite costs $0.10 per million input tokens and $0.40 per million output tokens, but Google limits 2.5 models to existing users. The cheapest option for new projects is Gemini 3.1 Flash-Lite at $0.25 input and $1.50 output per million tokens. Gemini 4 Argon costs $2 input and $10 output through November 30, then $4 and $20 from December 1.
Is there a free tier for the Gemini API?
Gemini 4 Hydrogen and some Flash models, including Gemini 3.8 Flash and 3.5 Flash-Lite, have a free tier. Gemini 4 Argon, Gemini 4 Carbon and Gemini 3.1 Pro Preview do not.
How much does Gemini 2.5 Pro cost?
In November 2026, Gemini 2.5 Pro costs $1.25 per million input tokens and $10 per million output tokens for requests up to 200,000 tokens. Above 200,000 tokens, the rates rise to $2.50 and $15. From December 1, 2026, calls to gemini-2.5-pro are scheduled to be served by Gemini 3.8 Flash and billed at its rates of $0.75 input and $3.75 output per million tokens through December 31.
When does Gemini 3.8 Flash pricing change?
Gemini 3.8 Flash costs $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026. On January 1, 2027, those rates double to $1.50 and $7.50.
How much does Gemini 3.8 Flash caching cost?
Through December 31, 2026, Gemini 3.8 Flash cached reads cost $0.075 per million tokens and explicit cache storage costs $0.50 per million tokens per hour. Implicit caching has no separate storage charge. On January 1, 2027, the cached read and explicit cache storage rates rise to $0.15 and $1.00.
What is the Gemini 3.8 Flash cache hit rate?
In Requesty's published rankings in November 2026, Gemini 3.8 Flash had a 53.6% cache hit rate and a blended cost of $0.47 per million tokens. Those figures describe the traffic in the rankings, so another workload may have different results.
Related reading

Start building with Requesty

One line of code. 600+ models. Full control.

Speak to founders