Gemini 3.6 Flash pricing: what it actually costs in 2026

Gemini 3.6 Flash costs $1.50 per 1M input tokens and $7.50 per 1M output tokens. That’s cheaper than the model it replaces. Google shipped the refresh on July 21, 2026, and for anyone watching an API bill, the headline is simple: the workhorse Flash tier got faster and less expensive at the same time.

Try Apidog today

This guide covers the numbers you need to budget: per-token rates, context caching costs, free-tier limits, and a worked example for estimating monthly spend before you ship. Every figure comes from Google’s Gemini API pricing docs and the launch announcement. For the full model overview, read what is Gemini 3.6 Flash.

One naming detail matters when comparing prices: the workhorse model moved to version 3.6, while the cheaper Flash-Lite tier remains at 3.5. They launched together, but the version numbers differ.

Gemini 3.6 Flash pricing at a glance

Item	Cost
Input	$1.50 per 1M tokens
Output (includes thinking tokens)	$7.50 per 1M tokens
Context caching, cached input read	$0.15 per 1M tokens
Context caching, storage	$1.00 per 1M tokens per hour
Free tier	Yes, via Google AI Studio (rate-limited)

Two details affect implementation budgets:

1. Thinking tokens are output tokens. Reasoning generated before the final answer is billed at the $7.50 per 1M output-token rate. There is no separate thinking-token line item.
2. Caching has read and storage charges. Cached input reads cost less, but storage is billed hourly. Caching only saves money when your application reuses the same prompt prefix or reference context frequently.

How it compares to Gemini 3.5 Flash

The output rate dropped from $9.00 per 1M tokens for Gemini 3.5 Flash to $7.50 per 1M tokens for Gemini 3.6 Flash. That is about a 17% lower output-token price.

Google also reports on the DeepMind Flash model page that Gemini 3.6 Flash produces roughly 17% fewer output tokens for the same task. Fewer reasoning steps and tool calls can reduce token use on multi-step workloads.

That means you can benefit from two cost reductions at once:

• Lower output-token pricing
• Fewer output tokens for the same task

If you are migrating an existing workload, compare your current usage with the Gemini 3.5 Flash pricing breakdown. For model capabilities and benchmarks, see Gemini 3.6 Flash vs 3.5 Flash.

What the free tier gets you

A free tier is available through Google AI Studio. You can call Gemini 3.6 Flash without a credit card for prototyping and testing.

Treat it as a development environment, not a production plan:

• The free tier is rate-limited.
• It is intended for prototyping and light testing.
• Google may use free-tier data to improve its products.
• Sensitive, proprietary, and customer data should use a paid key instead.

For setup details, see how to use Gemini 3.6 Flash for free. There is no unlimited free option, so plan a paid migration before production launch.

A worked cost example

Assume an agent uses:

• 2M input tokens per day
• 500K output tokens per day

This is a common pattern for a retrieval-heavy assistant that reads documents and returns relatively short answers.

Calculate the Gemini 3.6 Flash daily cost:

• Input: 2M × $1.50 per 1M = $3.00
• Output: 500K × $7.50 per 1M = $3.75
• Daily total: $6.75
• Monthly total over 30 days: $202.50

Now compare the output cost with Gemini 3.5 Flash.

If the same task emitted 600K output tokens on 3.5 Flash, a 17% reduction brings the 3.6 Flash output close to 500K tokens:

• Gemini 3.5 Flash: 600K × $9.00 per 1M = $5.40/day
• Gemini 3.6 Flash: 500K × $7.50 per 1M = $3.75/day

That is:

• $1.65 less per day
• About $49.50 less per month
• Roughly a 31% reduction in output spend

The lower rate and lower token count compound.

Flash-Lite is even cheaper

For simpler tasks, Gemini 3.5 Flash-Lite is the lower-cost tier:

Item	Cost
Input	$0.30 per 1M tokens
Output	$2.50 per 1M tokens
Cached input read	$0.03 per 1M tokens
Cache storage	$1.00 per 1M tokens per hour

Google reports Flash-Lite runs at around 350 output tokens per second.

For the same 2M-input, 500K-output workload:

• Input: 2M × $0.30 = $0.60/day
• Output: 500K × $2.50 per 1M = $1.25/day
• Daily total: $1.85
• Monthly total: about $55.50

That is roughly 70% cheaper than Gemini 3.6 Flash for identical token counts.

Use Flash-Lite for high-volume, low-complexity routes such as:

• Classification
• Extraction
• Routing
• Short structured responses

Use Gemini 3.6 Flash for harder multi-step reasoning. For more detail, read what is Gemini 3.5 Flash-Lite and Gemini 3.5 Flash-Lite vs 3.6 Flash.

How to control and measure your costs

Use these four levers to keep Gemini API costs predictable.

1. Cache repeated context

If requests share a long system prompt or the same reference documents, use context caching.

Cached reads cost $0.15 per 1M tokens instead of $1.50 per 1M tokens for standard input. However, cache storage costs $1.00 per 1M tokens per hour.

Caching is most useful when:

• Many requests reuse the same prompt prefix
• Reference documents are reused within the cache lifetime
• You can amortize storage cost across enough calls

Avoid caching one-off contexts. The storage charge can outweigh the saved input cost.

2. Trim prompts and cap output

Output costs five times more than input, but oversized prompts still add up at scale.

Practical controls include:

• Remove repeated boilerplate from prompts.
• Send only relevant retrieval chunks.
• Set a maximum output-token limit.
• Require structured output where possible.
• Monitor long-running agent loops and tool-call chains.

3. Route simple tasks to Flash-Lite

Do not use one model for every endpoint by default.

A practical routing strategy might look like this:

classification, extraction, routing  -> Gemini 3.5 Flash-Lite
multi-step reasoning, complex agents -> Gemini 3.6 Flash

Mixed routing can reduce spend while keeping higher-quality reasoning where it matters.

4. Record actual token usage

Do not budget from estimates alone. Gemini responses include usageMetadata, which reports the tokens used by a request.

Inspect and log those fields for every production call:

{
  "usageMetadata": {
    "promptTokenCount": 0,
    "candidatesTokenCount": 0,
    "thoughtsTokenCount": 0,
    "totalTokenCount": 0
  }
}

Use these values to calculate real costs:

input_cost  = promptTokenCount / 1_000_000 × 1.50
output_cost = (candidatesTokenCount + thoughtsTokenCount) / 1_000_000 × 7.50

The exact token counts returned by the API are the counts that drive the invoice.

An API client can help make this measurable. In Apidog, create a POST request for the Gemini API, store the key in an environment variable, and send requests using real prompts. Inspect usageMetadata in the response, then add assertions that flag unexpected output-token increases.

For example, you can test that a prompt update does not exceed a token budget:

const usage = pm.response.json().usageMetadata;

pm.test("Output stays under the token budget", () => {
  pm.expect(usage.candidatesTokenCount).to.be.below(1000);
});

You can also schedule API tests in Apidog to catch cost regressions before they reach production. To get started, download Apidog and point a request at the Gemini endpoint.

FAQ

How much does Gemini 3.6 Flash cost per 1M tokens?

It costs $1.50 per 1M input tokens and $7.50 per 1M output tokens. Context-cached reads cost $0.15 per 1M tokens, while storage costs $1.00 per 1M tokens per hour.

Does the output price include thinking tokens?

Yes. Thinking tokens are billed at the $7.50 per 1M output-token rate. There is no separate thinking-token charge, but additional reasoning increases total output usage.

Is there really a free tier?

Yes. Google AI Studio provides a rate-limited free tier for prototyping and testing. It is not intended for production traffic, and Google may use free-tier data to improve its products.

Is Gemini 3.6 Flash cheaper than 3.5 Flash?

Yes. The output price drops from $9.00 to $7.50 per 1M tokens, and Google reports that Gemini 3.6 Flash uses about 17% fewer output tokens for the same task. Together, those reductions can lower output spend by around 31%.

When should I use Flash-Lite instead?

Use Gemini 3.5 Flash-Lite for high-volume, low-complexity tasks such as classification, extraction, routing, and short structured replies. Use Gemini 3.6 Flash when the task requires more complex reasoning.

Gemini 3.6 Flash is a pricing update where both the rate and reported token use went down. Budget $1.50 per 1M input tokens and $7.50 per 1M output tokens, account for thinking tokens as output, and use caching and Flash-Lite routing where they fit.

For historical context, see the Gemini 3.0 API cost breakdown. Then measure real requests in Apidog so your budget is based on the token counts you actually pay for.