Mistral API costs are structured on a pay-as-you-go basis, with pricing determined by the specific model version and token volume. For high-reasoning tasks, Mistral Large 3 offers a standard on-demand rate of $0.50 per 1M input tokens. For high-volume, lower-complexity workflows, Mistral Small-3-1-24b-instruct-2503 provides a more economical entry point at $0.11 per 1M input tokens.
FinOps and engineering teams must distinguish between specific model versions as rates vary significantly. For example, the mistral-medium-2505 model carries a higher cost of $3.18 per 1M input tokens compared to the $0.64 rate for mistral-large-2512. Choosing the correct model tier is the primary lever for controlling API spend in production environments.
Which model tier provides the best value for your workflow?
Selecting a model requires balancing task complexity against per-token costs. Mistral Small-3-1-24b-instruct-2503 is the most budget-friendly for simple classification or extraction, costing $0.32 per 1M output tokens. Conversely, Mistral Medium-2505 is the most expensive tier at $9.50 per 1M output tokens, typically reserved for legacy integrations or specific reasoning requirements that justify the 30x price difference.
How does Mistral Large 3 compare to previous versions?
For teams scaling in the US, Mistral Large 3 provides a standard on-demand rate of $0.50 per 1M input tokens. This represents a slight cost reduction compared to the mistral-large-2512 pay-as-you-go input price of $0.64 per 1M tokens. Monitoring these version-specific price points is essential during model migration to ensure that performance gains do not result in unintended budget overruns.
Measuring the value of Mistral adoption requires moving beyond token tracking to net return. NetLift enables you to compare the $0.11–$3.18 input costs against the baseline cost of human labor or previous non-AI workflows. By quantifying the time saved per successful API call, you can determine if the higher cost of the Medium or Large models actually generates a superior financial return through increased accuracy and reduced manual oversight.
Sources
All pricing on this page comes from official vendor pages: