Mistral AI models — pricing & specs

All 13 Mistral models tracked by Mintok, with live $/M-token pricing, capability tiers and context windows. Sorted by capability, then price.

Cheapest
$0.10/M out
Most capable
tier 5/8
ModelTierClassOutput $/MInput $/MContext
Mistral Large 35/8Flagship$1.50$0.50—
Magistral Medium5/8Balanced$5$2—
Mistral Medium 3.55/8Balanced$7.50$1.50—
Mistral Large 25/8Flagship——131,072
Mixtral 8x22B4/8Balanced$6$265,536
Mistral NeMo3/8Nano$0.15$0.15—
Mistral Small 43/8Fast$0.60$0.15256,000
Mixtral 8x7B3/8Balanced$0.70$0.70—
Magistral Small3/8Fast$1.50$0.50—
Ministral 3 8B2/8Nano$0.15$0.15—
Ministral 3 14B2/8Nano$0.20$0.20—
Mistral 7B2/8Nano——32,768
Ministral 3 3B1/8Nano$0.10$0.10—

FAQ

How much do Mistral models cost?

Mistral models on the Mintok catalog range from $0.10 to $7.50 per million output tokens.

What is Mistral's cheapest model?

Mistral's cheapest model is Ministral 3 3B at $0.10 per million output tokens (tier 1/8).

What is Mistral's most capable model?

Mistral's highest-tier model on the catalog is Mistral Large 3 at capability tier 5/8, $1.50/M output.

How many Mistral models does Mintok track?

Mintok tracks 13 public Mistral models with live specs and pricing.

Compare Mistral on your workload

Benchmark Mistral models on your own prompts, or find the cheapest capable option across all vendors.

Prices maintained by Mintok's catalog watchers from public sources. Tiers per methodology. Not investment advice.