EN ▾
mistralapikey.com

Uncensored Mistral API alternative

Get API key
Base URL
https://api.mistralapikey.com/v1
Model
uncensored

Mistral API Pricing vs a Pay-Per-Token Alternative

For mistral api pricing, always read the vendor's current page, then compare on your own workload, not on headline rates. Here are four workloads priced at $0.15 per 1M input and $0.50 per 1M output tokens, plus a calculator you can reuse.

Updated

Two glass pipelines feeding coins into measuring jars, illustrating a comparison of per-token API pricing

How to compare any two per-token prices

Before the numbers, a rule: do not compare headline prices, compare costs for your workload. Every provider lists a price for input tokens and a price for output tokens, often per million. Your bill is the sum of both, weighted by how your app really behaves.

We will not quote Mistral's current rates here, because they change and the only reliable place to read them is Mistral's own pricing page. What we can do is give you exact costs for the alternative on this site, $0.15 per 1M input tokens and $0.50 per 1M output tokens, for four common workloads. Take those, look up the rates for the models you consider on the vendor's page, and put both in the small calculator below. Note that this service is not Mistral's API and not affiliated with it. It is a separate service with one open-weight model run by an independent provider.

Four workloads priced at $0.15 and $0.50

WorkloadTokens per request (in / out)Volume per monthCost per requestMonthly cost
Support chat1,200 / 25020,000 requests$0.000305$6.10
Classification400 / 20500,000 requests$0.00007$35.00
Document summaries30,000 / 8002,000 documents$0.0049$9.80
Story generation800 / 2,0005,000 generations$0.00112$5.60

Check the first row: 1,200 x $0.15 / 1M = $0.00018, and 250 x $0.50 / 1M = $0.000125, so $0.000305. Times 20,000 is $6.10. The pattern in the table matters more than any single figure. Classification looks cheap per call but dominates the monthly bill because of volume. Summaries are almost all input cost. Story generation is almost all output cost.

A calculator you can reuse

workloads = {
    "support chat": (20_000, 1_200, 250),
    "classification": (500_000, 400, 20),
    "summaries": (2_000, 30_000, 800),
    "stories": (5_000, 800, 2_000),
}

def cost(rate_in, rate_out):
    out = {}
    for name, (n, tin, tout) in workloads.items():
        out[name] = round(n * (tin * rate_in + tout * rate_out) / 1_000_000, 2)
    return out

print(cost(0.15, 0.50))
# {'support chat': 6.1, 'classification': 35.0, 'summaries': 9.8, 'stories': 5.6}

Call cost() a second time with the vendor's published input and output rates for the model you are considering. Rates in this function are dollars per 1M tokens. Replace the workload numbers with your own from logs, since your real token counts will not match round examples.

What the price does not tell you

A pay-per-token comparison leaves out the things that decide whether a model is usable. A cheaper model that needs two attempts per task costs double. A model that refuses your legitimate request costs you time even if the refused call is free. Consider:

  • Refusals. The model on this site is tuned to answer without content refusals for lawful adult use. That matters for fiction, roleplay and some research tasks, and it is irrelevant to others. A hard rule remains: no sexual content involving minors.
  • Quality for your task. Run your own prompts. A general benchmark says little about your use case, and we do not publish benchmark claims.
  • Context. The window here is 100,000 tokens for prompt plus completion.
  • Limits. A full key allows 8 concurrent requests and 300 per minute.
  • Ownership. One model, one provider. If you need several vendors' models, a different service fits better.

Cutting the bill without changing provider

Two changes on the workloads above show how much your own design matters, whatever rates you pay.

Batch the classification. Each classification request carries 300 tokens of instructions and 100 tokens of ticket text. Sending 10 tickets in one request needs 300 + 10 x 100 = 1,300 input tokens, and returns 10 x 20 = 200 output tokens. That costs 1,300 x $0.15 / 1M = $0.000195 plus 200 x $0.50 / 1M = $0.0001, or $0.000295 per call and 50,000 calls for 500,000 tickets. The month drops from $35.00 to $14.75. Use JSON mode (response_format: {"type": "json_object"}) so the ten answers come back in one parseable object.

Trim the summaries. Cutting each document from 30,000 to 20,000 input tokens through extraction of the relevant sections saves 10,000 x $0.15 / 1M = $0.0015 per document, so $3.00 across 2,000 documents. The month falls from $9.80 to $6.80.

Savings of this size are normal. They usually beat the difference between two providers' list prices, which is the point of pricing your own workload first.

Billing rules that change the total

Check these on every provider, including this one. Here, billing is prepaid and pay as you go, with no subscription. Credit never expires. Failed or refused requests are not charged. Top-ups are any whole amount from $20 to $500, and larger top-ups carry a bonus: +5% from $50, +10% from $100, +15% from $200. A $200 top-up therefore gives $230 of credit, which lowers the effective rates by about 13%. Pay by crypto (USDT on TRC20, USDC on Base). The pricing page shows the table.

Applied to the table above: the support chat workload costs $6.10 a month, so a $20 top-up covers more than three months of it.

Planning a budget and a top-up

Suppose one product runs all four workloads. The monthly total is $6.10 + $35.00 + $9.80 + $5.60 = $56.50. A $50 top-up gives $52.50 of credit, slightly under one month. A $100 top-up gives $110, which covers about 1.9 months. After the batching and trimming changes above, the total falls to $6.10 + $14.75 + $6.80 + $5.60 = $33.25, and the same $110 lasts more than three months.

Because credit never expires, the practical advice is to buy the amount that matches your confidence. Not sure yet? Start with $20 and watch the X-Balance-USD header on each response. Sure? A larger top-up earns the bonus. Either way, nothing renews automatically and nothing is lost if you pause for a month. The documentation covers every parameter used above.

Try it with your own prompts first

Sign up with Google or with email and password and claim the $0.50 trial credit with one click on the API key page. It needs no card, lasts 7 days and is limited to one per person. The trial key allows 2 concurrent requests and 60 per minute, which is enough for a test run.

curl https://api.mistralapikey.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"uncensored","messages":[{"role":"user","content":"Classify this ticket: refund request"}],"max_tokens":20}'

Read the usage block in the response and plug the real token counts into the calculator. For Python code and migration notes, see the article on the OpenAI-compatible chat API, and the page about the Mistral API key for background on keys.

Questions and answers

What are Mistral's current prices?

We do not quote them. They change, so read the vendor's own pricing page for the model you consider.

Is this service Mistral's API?

No. It is a separate service with one open-weight model run by an independent provider, not the official API of any vendor.

Which workload gets cheaper fastest with lower input rates?

Classification and summaries, since most of their tokens are input. Long generation depends mostly on the output rate.

Does credit expire?

No. Credit never expires, and failed or refused requests are not charged.

Your key is one form away

Create an account, copy the key, change the base URL. That is the whole setup.