Mistral token counter & API cost calculator
Paste any text to see how many tokens Mistral reads it as, and what it costs on each model. Counts update as you type.
| Model | Input $/M | Output $/M | Context | Per call | Per month |
|---|---|---|---|---|---|
|
Mistral Large 3
Mistral |
$0.50 | $1.50 | 128K | — | — |
|
Mistral Nemo
Mistral 3 hosts · up to ×1.2 dearer elsewhere |
$0.02 | $0.04 | 128K | — | — |
Rates are USD per million tokens, published list price, checked 2026-08-31. Batch and cached-input discounts are not applied. Confirm against the provider before committing a budget.
How Mistral counts tokens
Mistral uses its own tokeniser vocabulary, and this calculator applies one generic multiplier to every provider outside the big three rather than modelling it separately — so read the figure as a close approximation rather than an exact count.
The counter above is an estimate, not Mistral's own merge table. It models the behaviour that actually drives the number — a leading space merging into the word after it, digits grouping, characters outside the Latin range costing several times more — which is close enough to size a prompt and budget a workload, and deliberately not close enough to reconcile an invoice against.
The cheapest model on this whole site is a Mistral one
Nemo is priced roughly two orders of magnitude below the flagship models on this site, and it is a genuinely usable model rather than a toy. That changes the shape of the decision: on every other provider the question is which model to use, and here the first question is whether the job needs a large model at all. Classification, routing, tagging, extraction and short rewrites are the sort of work where a very cheap model finishes the task and the difference is invisible in the output but obvious on the bill. It is also open weights, so several hosts run it and the table marks the spread where we hold real rates for more than one.
What the 2 Mistral models cost
Take one representative call — a 2,000-token prompt and a 600-token reply, which is roughly two pages in and one page out — and run it 100 times a day.
the cheapest here
the dearest here
for identical work
on the dearest, vs $0.19 on the cheapest
Large 3 is the model to price when the work genuinely needs capability, and it sits well below the flagship rates of OpenAI and Anthropic while carrying a comparable context window. Between it and Nemo there is very little you cannot cover, which is why this page prices two models rather than a long ladder.
Which Mistral model to use
Nemo for volume, Large for the jobs that need real capability, and be honest about which one you are actually doing — most production traffic is the first. Mistral publishes other models, but only these two carry a rate we can confirm against mistral.ai itself, and this page deliberately prices nothing it cannot verify at source.
Context windows
The Mistral models here run from 128K tokens to 128K. That window is shared: your prompt, every earlier turn in the conversation, and the reply all have to fit inside it together. The table above marks a model in red when the text you have pasted plus the reply length you set would not fit — which is usually how people discover that a window is not as generous as the headline number suggests.
Questions
Is Mistral cheaper than OpenAI?
At the bottom of the range, dramatically — Nemo costs a small fraction of any GPT model. At the top the gap narrows. As always the honest comparison is cost per finished task rather than cost per token, since a cheaper model that needs three attempts is not cheaper.
Are Mistral models free to self-host?
Several are open weights, which means you can run them yourself or use any host that serves them, and the licence rather than the price is what governs commercial use. Running them still costs compute; free weights are not free inference.
Why are only two Mistral models listed here?
Because only these two carry a rate we could confirm against mistral.ai itself. Mistral publishes more models, but its pricing page renders most of the table in JavaScript, so the rates are not readable at source. Rather than copy figures from a third-party aggregator into a cost calculator, this page lists what it can verify and says so. A daily check re-tests every rate on this site against its vendor page.
Is this the official Mistral token count?
No, it is an estimate using a generic multiplier rather than the Mistral vocabulary. Close enough to budget with, not close enough to reconcile an invoice against.
Counting for a different provider
The same text costs a different number of tokens on every family, so if you are comparing providers, compare on your own text rather than on a rule of thumb.
Rates are USD per million tokens, published list price, checked 2026-08-31. Batch and cached-input discounts are not applied. Confirm against Mistral before committing a budget.
More free tools
All free, all instant, none of them need an account.