Gemini and Mistral plans compared with their API prices

Google's AI plans start at $4.99 a month in the US, and its Gemini API lists prices per million tokens updated on 16 October 2026. Mistral sells a Pro plan at $14.99 a month and API tokens priced separately. This post puts the published numbers next to each other for both vendors, as fetched on 18 October 2026.

Neither vendor publishes plan limits in tokens. Both use relative multipliers, so the comparison is about price shape, not a verified token allowance.

Google AI plans

From Google's US subscriptions page:

The page says Gemini app limits are compute based and depend on prompt complexity, features used and chat length. It adds that the limit refreshes every 5 hours until you reach a weekly limit, and you can buy AI credits to extend limits. No numeric cap is given.

Gemini API prices

From the Gemini API pricing page, Paid Tier, Standard, per million tokens. The page marks some prices with a change date, and we show both. The 3.1 Pro row is split by prompt length instead:

Batch and Flex are generally about half the Standard rate, and Priority is about 1.8x for most models, per the page. Storage for cached context is billed per million tokens per hour on models that support it, for example $0.50 on Gemini 3.8 Flash. The page also says that most text models have a free tier, and that free tier content may be used to improve Google's products. That is a data terms difference worth reading before you send source code.

Mistral

From mistral.ai/pricing:

For the API, the page shows per token prices for one model only: Mistral Large at $0.50 per million input tokens and $1.50 per million output tokens. It adds that batch processing takes 50 percent off and cached input can cut input cost by up to 90 percent. Prices for other Mistral models were not on the page we fetched, so we do not list them.

What you can and cannot compare

You can compare list prices per token across vendors, as long as you remember tokenizers differ, so the same text is a different token count on each. You can compare plan prices. You cannot compare plan allowances, because Google publishes multiples of Free, Mistral publishes "more", and neither gives tokens.

One thing Mistral's Pro page does give you is a dollar figure: $15 a month of API credits inside a $14.99 plan.

Try it yourself

Pick one model per vendor and price the same session. This uses a mid sized coding session of 1,000,000 input tokens (800,000 cached) and 20,000 output tokens.

models = {
    # name: (input, cached_input, output) USD per million tokens, from vendor pages on 18 Oct 2026
    "gemini-3.8-flash (now)":   (0.75, 0.075, 3.75),
    "gemini-3.8-flash (2027)":  (1.50, 0.15, 7.50),
    "mistral-large (no cache)": (0.50, 0.50, 1.50),
}
fresh, cached, out = 200_000, 800_000, 20_000
for name, (i, c, o) in models.items():
    print(name, round((fresh * i + cached * c + out * o) / 1e6, 3))

The Mistral row uses the plain input price for cached tokens because the page gives only "up to 90 percent", not a fixed cached price. Add the exact figure if you find it in your account.

If you route security tooling through any of these, Safeguard lets you bring your own model, and its MCP tool calls are unlimited.

Quick check

Sources

• Google AI plans (US), fetched 18 October 2026
• Gemini API pricing, page updated 16 October 2026, fetched 18 October 2026
• Mistral pricing, fetched 18 October 2026