Input (/M tokens)
Catalog row in Vendor License Metrics · Cited
- Aliases
- input tokens; per million tokens; /M tokens
- Definition
- “Most models are priced per million tokens, with both input (your prompts) and output (responses) counted separately.” (Mistral pricing FAQ, How is API pricing calculated?)
- Counting rules
- Input tokens (the prompt) are counted separately from output tokens and priced per million tokens per model. Batch processing reduces the price by 50%. Regional (EU) inference endpoints add 10% on supported models. Cached input tokens are a separate, lower rate.
- Floors / minimums
- None published. Pay-as-you-go against the price list unless an Order Form says otherwise (Commercial Terms §10.1).
- Applies to products
- Mistral Large 3
- Mistral Medium 3.5
- Mistral Small 4
- Ministral 3 14B
- Ministral 3 8B
- Ministral 3 3B
- Codestral
- Mistral Embed
- Codestral Embed
- Related metrics
- Product category