Ministral 3 API pricing

The ministral line is the only part of the Mistral range priced symmetrically: 0.1 per million tokens in and out for the 3B model, 0.15 for the 8B and 0.2 for the 14B, with output never costing more than input. The steps are small, half again from the first to the second, then a third more to the third, twice the cheapest at the top, which makes moving up a modest decision rather than a budget change. All three declare agentic and lightweight capabilities and none declares multimodal or multilingual support; that boundary rather than the price is what pushes a workload towards mistral-small-latest at 0.15 input and 0.6 output.

The rest of the range, and every rate Mistral AI publishes: Mistral AI API pricing