Mistral API Pricing

Compare all 13 Mistral models with detailed pricing, capabilities, and performance metrics.

All Models

Model NameCategoryInput PriceOutput PriceCached InputBatch InputContextMax OutputMultimodalFine-tuning
Mistral Large 3
Most capable Mistral model for complex tasks and analysis.
Flagship$0.500000$1.500000262.1K32.8KText
Magistral Medium
Mistral reasoning model for complex multi-step problems.
Flagship$2.000000$5.00000040K32.8KText
Mistral Medium 3.5
Latest Mistral mid-tier flagship. 128B dense model with configurable reasoning effort and 256K context. Released April 2026.
Flagship$1.500000$7.500000256K32.8KText
Mistral Medium 3.1
Balanced mid-tier Mistral model for production workloads.
Mid-Tier$0.400000$2.000000131.1K32.8KText
Codestral
Specialized model for code generation, completion, and analysis.
Mid-Tier$0.300000$0.900000256K32.8KText
Devstral 2
Agent-focused coding model for software development workflows with 256K max output.
Mid-Tier$0.400000$2.000000256K256KText
Mistral Small 4
Latest efficient Mistral model with 262K context.
Mid-Tier$0.150000$0.600000$0.015000262.1K16.4KText
Mistral Small 3.2
Ultra-efficient small model for high-volume applications.
Budget$0.075000$0.200000$0.010000131.1K32.8KText
Mistral Nemo
Lightweight open-weight model for simple tasks.
Budget$0.020000$0.040000131.1K32.8KText
Magistral Small
Compact Mistral reasoning model for multi-step problems.
Mid-Tier$0.500000$1.50000040K16.4KText
Ministral 14B
Compact edge model for on-device and high-volume applications.
Budget$0.200000$0.200000$0.020000262.1K16.4KText
Ministral 8B
Small edge model for lightweight tasks and fast inference.
Budget$0.150000$0.150000$0.015000262.1K16.4KText
Ministral 3B
Ultra-compact edge model for the cheapest inference.
Budget$0.040000$0.040000128K16.4KText

Prices in USD per 1 million tokens. Data last updated 5/12/2026.

Compare Mistral Models

Use the comparison tool to pit Mistral models against each other or models from other providers:

Open Pricing Comparison

Estimate Your Costs

Calculate how much you'll spend with Mistral's models based on your projected token usage.

Go to Cost Estimator

Frequently Asked Questions

How much does Mistral API cost?
Mistral offers a range of models with input pricing from $0.020000 to $2.00 per 1 million input tokens. Output pricing varies by model and ranges from $0.040000 to $7.50 per 1 million output tokens. Use the pricing comparison tool to see detailed breakdowns for all models.
Which Mistral model is cheapest?
The Mistral Nemo is Mistral's most cost-effective option at $0.020000 per 1M input tokens. It is text-only. Check the pricing table above to compare all available models or use the cost estimator tool for your specific use case.
Does Mistral offer batch API pricing?
Currently, Mistral does not appear to offer batch API pricing. However, pricing and features change frequently. Check Mistral's official pricing page for the latest offerings.