mistral.yml

# yaml-language-server: $schema=.schema.json
name: Mistral
id: mistral
pricing_urls:
  - https://mistral.ai/pricing#api-pricing
api_pattern: 'https://api\.mistral\.ai'
provider_match:
  starts_with: mistral
model_match:
  regex: (?:mi|code|dev|magi|mini)stral

# https://github.com/mistralai/client-python/blob/v1.9.3/src/mistralai/models/usageinfo.py
extractors:
  - root: usage
    mappings:
      - path: prompt_tokens
        dest: input_tokens
      - path: completion_tokens
        dest: output_tokens

models:
  - id: codestral
    name: Codestral
    description: >-
      Mistral's cutting-edge language model for coding. Codestral specializes in low-latency, high-frequency tasks such as
      fill-in-the-middle (FIM), code correction and test generation.
    match:
      or:
        - equals: codestral-latest
        - equals: codestral-2501
    prices_checked: 2025-07-04
    prices:
      input_mtok: 0.3
      output_mtok: 0.9

  - id: devstral-small
    name: Devstral Small
    description: >-
      Devstral-Small-2505 is a 24B parameter agentic LLM fine-tuned from Mistral-Small-3.1, jointly developed by Mistral AI
      and All Hands AI for advanced software engineering tasks. It is optimized for
      codebase exploration, multi-file editing, and integration into coding agents, achieving state-of-the-art results on
      SWE-Bench Verified (46.8%).
    match:
      equals: devstral-small
    prices:
      input_mtok: 0.06
      output_mtok: 0.12

  - id: devstral-small:free
    name: Devstral Small (free)
    description: >-
      Devstral-Small-2505 is a 24B parameter agentic LLM fine-tuned from Mistral-Small-3.1, jointly developed by Mistral AI
      and All Hands AI for advanced software engineering tasks. It is optimized for
      codebase exploration, multi-file editing, and integration into coding agents, achieving state-of-the-art results on
      SWE-Bench Verified (46.8%).
    match:
      equals: devstral-small:free
    prices: {}

  - id: magistral-medium
    name: Magistral Medium
    description: >-
      Magistral is Mistral's first reasoning model. It is ideal for general purpose use requiring longer thought processing
      and better accuracy than with non-reasoning LLMs. From legal research and financial
      forecasting to software development and creative storytelling — this model solves multi-step challenges where transparency
      and precision are critical.
    match:
      or:
        - starts_with: magistral-medium
    prices_checked: 2025-08-12
    prices:
      input_mtok: 2
      output_mtok: 5

  - id: magistral-small
    name: Magistral Small
    description: >-
      Magistral Small is a 24B parameter instruction-tuned model based on Mistral-Small-3.1 (2503), enhanced through supervised
      fine-tuning on traces from Magistral Medium and further refined via reinforcement
      learning. It is optimized for reasoning and supports a wide multilingual range, including over 20 languages.
    match:
      starts_with: magistral-small-
    prices_checked: 2025-08-12
    prices:
      input_mtok: 0.5
      output_mtok: 1.5

  - id: ministral-3b
    name: Ministral 3B
    description: >-
      Ministral 3B is a 3B parameter model optimized for on-device and edge computing. It excels in knowledge, commonsense
      reasoning, and function-calling, outperforming larger models like Mistral 7B on
      most benchmarks. Supporting up to 128k context length, it's ideal for orchestrating agentic workflows and specialist
      tasks with efficient inference.
    match:
      equals: ministral-3b
    prices:
      input_mtok: 0.04
      output_mtok: 0.04

  - id: ministral-8b
    name: Ministral 8B 24.10
    description: >-
      Ministral 8B is an 8B parameter model featuring a unique interleaved sliding-window attention pattern for faster, memory-efficient
      inference. Designed for edge use cases, it supports up to 128k context
      length and excels in knowledge and reasoning tasks. It outperforms peers in the sub-10B category, making it perfect
      for low-latency, privacy-first applications.
    match:
      starts_with: ministral-8b
    prices_checked: 2025-08-12
    prices:
      input_mtok: 0.1
      output_mtok: 1

  - id: mistral-7b
    name: Mistral 7B
    match:
      or:
        - equals: mistral-7b
        - equals: open-mistral-7b
    prices_checked: 2025-08-12
    prices:
      input_mtok: 0.25
      output_mtok: 0.25

  - id: mistral-embed
    match:
      equals: mistral-embed
    prices:
      input_mtok: 0.1
      output_mtok: 0.1

  - id: mistral-large
    name: Mistral Large
    description: >-
      This is Mistral AI's flagship model, Mistral Large 2 (version `mistral-large-2407`). It's a proprietary weights-available
      model and excels at reasoning, code, JSON, chat, and more. Read the launch
      announcement here.
    match:
      or:
        - equals: mistral-large
        - equals: mistral-large-latest
        - equals: mistral-large-2407
        - equals: mistral-large-2411
    prices_checked: 2025-07-04
    prices:
      input_mtok: 2
      output_mtok: 6

  - id: mistral-medium-3
    name: Mistral Medium 3
    description: >-
      Mistral Medium 3 is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities
      at significantly reduced operational cost. It balances state-of-the-art reasoning
      and multimodal performance with 8× lower cost compared to traditional large models, making it suitable for scalable
      deployments across professional and industrial use cases.
    match:
      starts_with: mistral-medium
    prices_checked: 2025-08-12
    prices:
      input_mtok: 0.4
      output_mtok: 2

  - id: mistral-nemo
    name: Mistral NeMo
    description: >-
      A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA.
    match:
      or:
        - equals: mistral-nemo
        - equals: open-mistral-nemo
    prices_checked: 2025-07-04
    prices:
      input_mtok: 0.15
      output_mtok: 0.15

  - id: mistral-nemo:free
    name: Mistral Nemo (free)
    description: >-
      A 12B parameter model with a 128k token context length built by Mistral in collaboration with NVIDIA.
    match:
      equals: mistral-nemo:free
    prices: {}

  - id: mistral-saba
    name: Mistral Saba
    description: >-
      Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering
      accurate and contextually relevant responses while maintaining efficient performance.
      Trained on curated regional datasets, it supports multiple Indian-origin languages—including Tamil and Malayalam—alongside
      Arabic. This makes it a versatile option for a range of regional and multilingual
      applications. Read more at the blog post here
    match:
      or:
        - equals: mistral-saba
        - equals: mistral-saba-latest
    prices_checked: 2025-07-04
    prices:
      input_mtok: 0.2
      output_mtok: 0.6

  - id: mistral-small-24b-instruct-2501
    name: Mistral Small 3
    description: >-
      Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released
      under the Apache 2.0 license, it features both pre-trained and instruction-tuned
      versions designed for efficient local deployment.
    match:
      equals: mistral-small-24b-instruct-2501
    prices_checked: 2025-07-05
    price_comments: Can't find pricing on this model, so just trusting open router
    prices:
      input_mtok: 0.05
      output_mtok: 0.08

  - id: mistral-small-24b-instruct-2501:free
    name: Mistral Small 3 (free)
    description: >-
      Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released
      under the Apache 2.0 license, it features both pre-trained and instruction-tuned
      versions designed for efficient local deployment.
    match:
      equals: mistral-small-24b-instruct-2501:free
    prices: {}

  - id: mistral-small-latest
    description: SOTA. Multimodal. Multilingual. Apache 2.0.
    name: Mistral Small 3.2
    match:
      equals: mistral-small-latest
    prices_checked: 2025-07-04
    prices:
      input_mtok: 0.1
      output_mtok: 0.3

  - id: mistral-tiny
    name: Mistral Tiny
    description: >-
      Note: This model is being deprecated. Recommended replacement is the newer Ministral 8B
    match:
      equals: mistral-tiny
    deprecated: true
    prices:
      input_mtok: 0.25
      output_mtok: 0.25

  - id: mixtral-8x22b-instruct
    name: Mixtral 8x22B Instruct
    description: >-
      Mistral's official instruct fine-tuned version of Mixtral 8x22B. It uses 39B active parameters out of 141B, offering
      unparalleled cost efficiency for its size. Its strengths include:

      - strong math, coding, and reasoning

      - large context length (64k)

      - fluency in English, French, Italian, German, and Spanish
    match:
      equals: mixtral-8x22b-instruct
    prices:
      input_mtok: 0.9
      output_mtok: 0.9

  - id: mixtral-8x7b
    name: Mixtral 8x7B
    match:
      or:
        - starts_with: mixtral-8x7b
        - equals: open-mixtral-8x7b
    prices_checked: 2025-08-12
    prices:
      input_mtok: 0.7
      output_mtok: 0.7

  - id: pixtral-12b
    name: Pixtral 12B
    description: >-
      The first multi-modal, text+image-to-text model from Mistral AI. Its weights were launched via torrent: https://x.com/mistralai/status/1833758285167722836.
    match:
      or:
        - equals: pixtral-12b
        - equals: pixtral-12b-latest
    prices_checked: 2025-07-04
    prices:
      input_mtok: 0.15
      output_mtok: 0.15

  - id: pixtral-large
    name: Pixtral Large 2411
    description: >-
      Pixtral Large is a 124B parameter, open-weight, multimodal model built on top of Mistral Large 2. The model is able
      to understand documents, charts and natural images.
    match:
      or:
        - equals: pixtral-large-latest
        - equals: pixtral-large-2411
    prices:
      input_mtok: 2
      output_mtok: 6