Gemini 3.1 Pro API relay: $1.12 and $6.49 at CheapRelay

01 / What does Gemini 3.1 Pro cost on CheapRelay?

On the CheapRelay price list, Gemini 3.1 Pro is $1.12 per 1,000,000 input tokens, $6.49 per 1,000,000 output tokens, and $0.1125 per 1,000,000 cache-read tokens. The model id you enter is gemini-3.1-pro.

That is three lines on the bill, and each line has its own rate. Input is what you send to the model. Output is what the model sends back. Cache read is the discounted rate for content the service can reuse from a cached prompt. You are not buying a fixed bundle of traffic, so the only number that matters for a given request is how many real tokens it uses.

Example: a request with 100,000 input tokens, 20,000 output tokens and 50,000 cache-read tokens costs 0.1 × $1.12 = $0.112 for input, 0.02 × $6.49 = $0.1298 for output, and 0.05 × $0.1125 = $0.005625 for cache reads. The example total is $0.247425.

There is no monthly fee. CheapRelay uses a prepaid balance, you pay per real token, and the balance does not expire. Every new account also receives $0.09 in sign-up credit, usable on any model, including Gemini 3.1 Pro.

02 / How do I set up Gemini 3.1 Pro without guessing?

Gemini models on CheapRelay use the native protocol only, under the /v1beta path, and they run through Gemini CLI. They are not OpenAI-compatible, so an OpenAI-style client setup is the wrong route for this model.

Example setup: create an account, point Gemini CLI at the CheapRelay API base, use the native Gemini protocol with the /v1beta path, and enter gemini-3.1-pro as the model. The model id matters because the price list publishes Gemini 3.1 Pro under that exact id.

Two operational limits are worth knowing before you build a workflow around the key. Each key allows 120 calls per minute, and each account can run 8 calls at once. If a request returns HTTP 402, the account balance is at or below zero, so top up the prepaid balance and retry.

Because billing follows real token usage, the same model can produce different line totals from one run to the next. That is normal for token-based pricing. If you want the cost of a request before you send it, calculate it with the three per-million rates above and multiply each one by the fraction of a million tokens you expect.

03 / How does Gemini 3.1 Pro compare with other Gemini models?

Gemini 3.1 Pro sits above the two other Gemini models on the CheapRelay price list. Gemini 3.8 Flash is $0.95 input, $4.28 output and $0.0944 cache read. Gemini 2.5 Flash is $0.1778 input, $1.42 output and $0.0181 cache read. Gemini 3.1 Pro is $1.12 input, $6.49 output and $0.1125 cache read.

Example comparison, using 100,000 input tokens and 20,000 output tokens: Gemini 3.1 Pro costs 0.1 × $1.12 + 0.02 × $6.49 = $0.112 + $0.1298 = $0.2418. Gemini 3.8 Flash costs 0.1 × $0.95 + 0.02 × $4.28 = $0.095 + $0.0856 = $0.1806. In this example, Gemini 3.8 Flash is $0.0612 less than Gemini 3.1 Pro.

Example comparison against Gemini 2.5 Flash with the same token counts: 0.1 × $0.1778 + 0.02 × $1.42 = $0.01778 + $0.0284 = $0.04618. That example is $0.19562 below the Gemini 3.1 Pro figure of $0.2418.

So the arithmetic tells you where to spend. If a task is mostly short prompts with modest replies, the lower Gemini 2.5 Flash rates keep the example bill small. If you need Gemini 3.1 Pro specifically, you know the three rates in advance and can budget against them. CheapRelay publishes the price list with those figures side by side, so the choice is a calculation rather than a guess.

04 / How does the Gemini 3.1 Pro rate compare with official list prices elsewhere?

Gemini 3.1 Pro does not have an official list price on the CheapRelay price list, so there is no official figure to subtract from its $1.12 input and $6.49 output rates. The honest comparison is with the official list prices that CheapRelay does publish for Claude and GPT models.

Example: Claude Sonnet 5.5 has an official list price of $2.00 input and $10.00 output, while its CheapRelay rate is $0.8903 input and $4.45 output. For 100,000 input tokens and 20,000 output tokens, the official list example is 0.1 × $2.00 + 0.02 × $10.00 = $0.20 + $0.20 = $0.40. The CheapRelay example is 0.1 × $0.8903 + 0.02 × $4.45 = $0.08903 + $0.089 = $0.17803, a difference of $0.22197 in this example.

Example: GPT-6.1 Sol has an official list price of $2.00 input and $10.00 output, and a CheapRelay rate of $0.1486 input and $0.7375 output. The published output gap is 93% under list. For 100,000 input and 20,000 output tokens, the official list example is $0.40, while the CheapRelay example is 0.1 × $0.1486 + 0.02 × $0.7375 = $0.01486 + $0.01475 = $0.02961.

The point is the method, not a borrowed number. CheapRelay shows the published gap to official list prices where those list prices exist, and it shows every Gemini rate on the same price list. You can run the same multiplication for Gemini 3.1 Pro, Gemini 3.8 Flash or Gemini 2.5 Flash and see exactly what your usage costs.

05 / What happens to your balance, credit and rewards?

Your prepaid balance is the working capital for every request. You top it up, CheapRelay deducts the real token cost, and unused balance stays available because it does not expire. If you want a refund on unused balance, request it within 12 days of the top-up. The refunded amount is the unused balance minus what was spent and minus the payment fee.

Example: top up $10, spend $4.26, and request a refund within 12 days. The unused balance is $10 - $4.26 = $5.74. After the payment fee is deducted, the refund is $5.74 minus that fee.

The referral program also uses published percentages. You receive 20% of what a directly invited user pays, plus 6% from a second-level user. Those payments land in a withdrawable rewards balance, and the smallest withdrawal is ¥20, paid through Alipay.

Put together, the setup is short: use Gemini CLI with the native protocol and the /v1beta path, enter gemini-3.1-pro, and watch three rates on the bill at $1.12 input, $6.49 output and $0.1125 cache read per million tokens. CheapRelay keeps the model list, the price list and the billing rules in plain numbers, so you can check the arithmetic before you spend.