mistral.yml

# yaml-language-server: $schema=.schema.json
name: Mistral
id: mistral
pricing_urls:
  - https://mistral.ai/pricing/api
api_pattern: 'https://api\.mistral\.ai'
price_comments: >-
  Batch API usage is charged at 50% of the standard rates for every model ("We offer a 50% discount on batch API").
  Batch rates for chat models, OCR 4.1 pages and Voxtral Mini Transcribe 2 are read from the Batch pricing mode at
  https://docs.mistral.ai/inference/pricing; OCR annotated pages, OCR 3, Voxtral Small and Mistral Embed apply the
  documented 50% discount to their current standard rates. Only models Mistral still serves carry a batch variant.
  https://docs.mistral.ai/capabilities/batch/
provider_match:
  starts_with: mistral
model_match:
  regex: "^(?![^/]+/)(?:labs-)?(?:(?:mi|code|dev|magi|mini)stral|mixtral|pixtral|voxtral|open-(?:mistral|mixtral))"

# 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: [prompt_tokens_details, cached_tokens]
        dest: cache_read_tokens
        required: false
      - path: [completion_tokens_details, reasoning_tokens]
        dest: output_reasoning_tokens
        required: false
      - path: completion_tokens
        dest: output_tokens
  # https://docs.mistral.ai/api/endpoint/ocr
  - api_flavor: ocr
    root: usage_info
    mappings:
      - path: pages_processed
        dest: input_document_pages
  - api_flavor: ocr_annotated
    root: usage_info
    mappings:
      - path: pages_processed
        dest: input_document_pages
      - path: pages_processed
        dest: input_annotated_document_pages

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
    context_window: 256000
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.3
      output_mtok: 0.9
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.15
          output_mtok: 0.45

  - id: codestral-2508
    name: Codestral 2508
    description: >-
      Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency
      tasks such as fill-in-the-middle (FIM), code correction and test generation.
    match:
      equals: codestral-2508
    context_window: 256000
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.3
      cache_read_mtok: 0.03
      output_mtok: 0.9
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.15
          cache_read_mtok: 0.015
          output_mtok: 0.45

  - id: codestral-embed-2505
    name: Codestral Embed 2505
    description: >-
      Mistral's code-specialized embedding model for code retrieval and semantic search.
    match:
      or:
        - equals: codestral-embed-2505
        - equals: codestral-embed
    context_window: 8000
    prices_checked: 2026-10-15
    price_comments: >-
      Rates from https://docs.mistral.ai/inference/pricing (Codestral Embed: input $0.15 / 1M tokens,
      cached input $0.015 / 1M tokens) and https://docs.mistral.ai/models/codestral-embed-25-05
      (embedding model, 8k context; embeddings bill input tokens only).
    prices:
      input_mtok: 0.15
      cache_read_mtok: 0.015

  - id: devstral-2512
    name: Devstral 2 2512
    description: >-
      Devstral 2 is a state-of-the-art open-source model by Mistral AI specializing in agentic coding. It is a 123B-parameter
      dense transformer model supporting a 256K context window.
    match:
      equals: devstral-2512
    context_window: 262144
    deprecated: true
    prices_checked: 2026-10-12
    price_comments: >-
      Deprecated 2026-05-22. Mistral recommends Mistral Medium 3.5 for new integrations.
      https://docs.mistral.ai/models/devstral-2-25-12
    prices:
      input_mtok: 0.4
      cache_read_mtok: 0.04
      output_mtok: 2

  - 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:
      or:
        - equals: devstral-small
        - equals: devstral-small-2505
        - equals: devstral-small-2507
        - equals: devstral-small-latest
        - equals: labs-devstral-small-2512
    # Every Devstral Small generation is deprecated or retired: 1.0 (devstral-small-2505) retired 2025-11-30,
    # 1.1 (devstral-small-2507) retired 2026-05-31, Small 2 (labs-devstral-small-2512, devstral-small-latest)
    # retired 2026-03-31.
    deprecated: true
    prices_checked: 2026-08-27
    price_comments: >-
      The previous $0.06/$0.12 rates were not Mistral's; every Devstral Small generation is documented at $0.10/$0.30.
      Ref: https://github.com/mistralai/platform-docs-public/tree/main/src/schema/models/models (devstral-small-*.ts)
    prices:
      input_mtok: 0.1
      output_mtok: 0.3

  - 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
    deprecated: true
    prices_checked: 2026-10-12
    price_comments: >-
      Magistral Medium 1.2 was deprecated 2026-05-22. Mistral recommends Mistral Medium 3.5.
      https://docs.mistral.ai/models/magistral-medium-1-2-25-09
      No context_window: this record matches versions with different context limits.
    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-
    deprecated: true
    prices_checked: 2026-10-12
    price_comments: >-
      Magistral Small 1.2 was deprecated 2026-04-30. Mistral recommends Mistral Small 4.
      https://docs.mistral.ai/models/magistral-small-1-2-25-09
      No context_window: this record matches versions with different context limits.
    prices:
      input_mtok: 0.5
      output_mtok: 1.5

  - id: ministral-14b-2512
    name: Ministral 3 14B 2512
    description: >-
      The largest model in the Ministral 3 family, Ministral 3 14B offers frontier capabilities and performance comparable
      to its larger Mistral Small 3.2 24B counterpart.
    match:
      equals: ministral-14b-2512
    context_window: 262144
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.2
      cache_read_mtok: 0.02
      output_mtok: 0.2
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.1
          cache_read_mtok: 0.01
          output_mtok: 0.1

  - 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-3b-2512
    name: Ministral 3 3B 2512
    description: >-
      The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision
      capabilities.
    match:
      equals: ministral-3b-2512
    context_window: 131072
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.1
      cache_read_mtok: 0.01
      output_mtok: 0.1
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.05
          cache_read_mtok: 0.005
          output_mtok: 0.05

  - 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:
      or:
        - equals: ministral-8b
        - equals: ministral-8b-2410
    prices_checked: 2026-08-24
    price_comments: >-
      The previous $1 output price was incorrect; Mistral documented $0.10 from launch.
      Ref: https://github.com/mistralai/platform-docs-public/blob/4422b455194a/src/schema/models/models/ministral-8b-24-1.ts
    prices:
      input_mtok: 0.1
      output_mtok: 0.1

  - id: ministral-8b-2512
    name: Ministral 3 8B 2512
    description: >-
      Ministral 3 8B is an efficient text and vision model for edge deployment.
    match:
      equals: ministral-8b-2512
    context_window: 262144
    prices_checked: 2026-10-14
    price_comments: >-
      Ref: https://mistral.ai/pricing/api
    prices:
      input_mtok: 0.15
      cache_read_mtok: 0.015
      output_mtok: 0.15
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.075
          cache_read_mtok: 0.0075
          output_mtok: 0.075

  - id: ministral-8b-latest
    name: Ministral 8B Latest
    match:
      equals: ministral-8b-latest
    # No context_window: this alias pointed to Ministral 8B 24.10 (128k) until 2025-12-02,
    # now to Ministral 3 8B (262,144) — no single value is true for every request it matches
    prices_checked: 2026-10-14
    price_comments: >-
      The latest alias moved from Ministral 8B 24.10 to Ministral 3 8B on 2025-12-02.
      Ref: https://github.com/mistralai/platform-docs-public/commit/4975b09514f95978cfeeea814562000348548107
    prices:
      - prices:
          input_mtok: 0.1
          output_mtok: 0.1
      - constraint:
          start_date: 2025-12-02
        prices:
          input_mtok: 0.15
          cache_read_mtok: 0.015
          output_mtok: 0.15
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.075
          cache_read_mtok: 0.0075
          output_mtok: 0.075

  - 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
    context_window: 8192
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.1
      output_mtok: 0.1
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.05
          output_mtok: 0.05

  - 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
    # No context_window: this record also matches the retired mistral-large-2407/2411 aliases
    # (131,072) vs mistral-large-latest's 262,144 — no single value is true for every ref it matches
    prices:
      input_mtok: 2
      output_mtok: 6

  - id: mistral-large-2512
    name: Mistral Large 3 2512
    description: >-
      Mistral Large 3 2512 is Mistral's most capable model to date, featuring a sparse mixture-of-experts architecture with
      41B active parameters (675B total), and released under the Apache 2.0 license.
    match:
      equals: mistral-large-2512
    context_window: 262144
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.5
      cache_read_mtok: 0.05
      output_mtok: 1.5
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.25
          cache_read_mtok: 0.025
          output_mtok: 0.75

  - id: mistral-large-4
    name: Mistral Large 4
    match:
      or:
        - equals: mistral-large-4
        - equals: mistral-large-4-0
    context_window: 1048576
    prices_checked: 2026-10-14
    price_comments: >-
      Current launch promotional rates are $0.68 input, $0.07 cached input, and $2.09 output per million tokens. The
      published list rates are $1.36/$0.14/$4.18. The October 6 changelog says the promotion lasts two weeks but does not
      state an exact effective end date, so no future price change is scheduled here. The batch variant is half the
      promotional rate; when the list rates return, prepend a dated batch variant at $0.68/$0.07/$2.09.
      https://docs.mistral.ai/models/mistral-large-4-0
      https://docs.mistral.ai/inference/pricing
      https://docs.mistral.ai/resources/changelogs
    prices:
      input_mtok: 0.68
      cache_read_mtok: 0.07
      output_mtok: 2.09
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.34
          cache_read_mtok: 0.035
          output_mtok: 1.045

  - id: mistral-medium-2312
    name: Mistral Medium 1
    description: Mistral's first enterprise-grade model, released in December 2023.
    match:
      equals: mistral-medium-2312
    prices_checked: 2026-08-24
    price_comments: >-
      Retired on 2025-06-16; retained for historical usage records at its original $2.70/$8.10 per MTok rates.
      Ref: https://github.com/mistralai/platform-docs-public/blob/222f4ba9114f84ce9f2d718b9a12fbac1e527f4b/src/schema/models/models/mistral-medium-1-0-23-12.ts
    prices:
      input_mtok: 2.7
      output_mtok: 8.1

  - id: mistral-medium-3-1
    name: Mistral Medium 3 and 3.1
    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:
      or:
        - equals: mistral-medium
        - equals: mistral-medium-2505
        - equals: mistral-medium-2508
    prices_checked: 2026-08-24
    price_comments: >-
      Mistral Medium 3 and 3.1 retain their original $0.40/$2 per MTok rates.
      Ref: https://github.com/mistralai/platform-docs-public/blob/222f4ba9114f84ce9f2d718b9a12fbac1e527f4b/src/schema/models/models/mistral-medium-3-1-25-08.ts
    # No context_window: live aliases disagree — mistral-medium reports 262,144 while
    # mistral-medium-2505/2508 report 131,072 — no single value is true for every ref it matches
    prices:
      input_mtok: 0.4
      cache_read_mtok: 0.04
      output_mtok: 2

  - id: mistral-medium-3-5
    name: Mistral Medium 3.5
    description: >-
      Mistral Medium 3.5 is a frontier-class multimodal model optimized for agentic and coding use cases.
    match:
      or:
        - equals: mistral-medium-3.5
        - equals: mistral-medium-3-5
        - equals: mistral-medium-3
        - equals: mistral-medium-2604
    context_window: 262144
    prices_checked: 2026-10-14
    price_comments: >-
      Ref: https://mistral.ai/pricing/api
    prices:
      input_mtok: 1.5
      cache_read_mtok: 0.15
      output_mtok: 7.5
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.75
          cache_read_mtok: 0.075
          output_mtok: 3.75

  - id: mistral-medium-latest
    name: Mistral Medium Latest
    match:
      equals: mistral-medium-latest
    # No context_window: this alias pointed to the 128k mistral-medium-2505 before repointing
    # to the 262,144 Medium 3.5 — no single value is true for every request it matches
    prices_checked: 2026-10-14
    price_comments: >-
      The latest alias moved from Mistral Medium 3.1 to Mistral Medium 3.5 on 2026-06-16.
      Ref: https://github.com/mistralai/platform-docs-public/commit/cc58c1186b1ca8ad65658f5dfd3ebd29de778c7f
    prices:
      - prices:
          input_mtok: 0.4
          cache_read_mtok: 0.04
          output_mtok: 2
      - constraint:
          start_date: 2026-06-16
        prices:
          input_mtok: 1.5
          cache_read_mtok: 0.15
          output_mtok: 7.5
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.75
          cache_read_mtok: 0.075
          output_mtok: 3.75

  - 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-ocr-2503
    name: Mistral OCR
    match:
      or:
        - equals: mistral-ocr-2503
        - equals: mistral-ocr-2503-completion
    deprecated: true
    prices_checked: 2026-08-25
    price_comments: >-
      Mistral launched OCR at 1,000 pages per US dollar:
      https://mistral.ai/news/mistral-ocr/
    prices:
      input_document_kpages: 1

  - id: mistral-ocr-2505
    name: Mistral OCR 2
    match:
      or:
        - equals: mistral-ocr-2505
        - equals: mistral-ocr-2505-completion
    deprecated: true
    prices_checked: 2026-08-25
    price_comments: >-
      OCR 2 retained the original OCR page price and introduced structured annotations on May 22, 2025:
      https://docs.mistral.ai/resources/changelogs
    prices:
      input_document_kpages: 1
      input_annotated_document_kpages: 3

  - id: mistral-ocr-2512
    name: Mistral OCR 3
    match:
      or:
        - equals: mistral-ocr-2512
        - equals: mistral-ocr-2512-completion
    context_window: 16384
    prices_checked: 2026-10-14
    price_comments: https://docs.mistral.ai/models/ocr-3-25-12
    prices:
      input_document_kpages: 2
      input_annotated_document_kpages: 3
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_document_kpages: 1
          input_annotated_document_kpages: 1.5

  - id: mistral-ocr-4-0
    name: Mistral OCR 4.0
    match:
      or:
        - equals: mistral-ocr-4-0
        - equals: mistral-ocr-4-0-completion
    context_window: 16384
    deprecated: true
    prices_checked: 2026-10-12
    price_comments: >-
      Deprecated 2026-09-29. Mistral recommends OCR 4.1 for new integrations.
      https://docs.mistral.ai/models/ocr-4-0
    prices:
      input_document_kpages: 4
      input_annotated_document_kpages: 5

  - id: mistral-ocr-4-1
    name: Mistral OCR 4.1
    match:
      or:
        - equals: mistral-ocr-4
        - equals: mistral-ocr-4-completion
        - equals: mistral-ocr-4-1
        - equals: mistral-ocr-4-1-completion
    context_window: 16384
    prices_checked: 2026-10-14
    price_comments: https://docs.mistral.ai/models/ocr-4-1
    prices:
      input_document_kpages: 4
      input_annotated_document_kpages: 5
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_document_kpages: 2
          input_annotated_document_kpages: 2.5

  - id: mistral-ocr-latest
    name: Mistral OCR Latest
    match:
      or:
        - equals: mistral-ocr-latest
        - equals: mistral-ocr-latest-completion
    # No context_window: this alias pointed to the 40k mistral-ocr-2505 before today's
    # 16,384 OCR 4 — no single value is true for every request it matches
    prices_checked: 2026-10-14
    price_comments: >-
      The latest alias moved to OCR 3 on December 18, 2025, OCR 4.0 on June 23, 2026, and OCR 4.1 on July 16, 2026.
      OCR 4.1 kept the OCR 4.0 rates: https://docs.mistral.ai/resources/changelogs
    prices:
      - prices:
          input_document_kpages: 1
      - constraint:
          start_date: 2025-05-22
        prices:
          input_document_kpages: 1
          input_annotated_document_kpages: 3
      - constraint:
          start_date: 2025-12-18
        prices:
          input_document_kpages: 2
          input_annotated_document_kpages: 3
      - constraint:
          start_date: 2026-06-23
        prices:
          input_document_kpages: 4
          input_annotated_document_kpages: 5
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_document_kpages: 2
          input_annotated_document_kpages: 2.5

  - 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-2603
    name: Mistral Small 4
    description: >-
      Mistral Small 4 is the next major release in the Mistral Small family, unifying the capabilities of several flagship
      Mistral models into a single system.
    match:
      equals: mistral-small-2603
    context_window: 262144
    prices_checked: 2026-10-14
    prices:
      input_mtok: 0.15
      cache_read_mtok: 0.015
      output_mtok: 0.6
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.075
          cache_read_mtok: 0.0075
          output_mtok: 0.3

  - id: mistral-small-3.1-24b-instruct
    name: Mistral Small 3.1 24B
    description: >-
      Mistral Small 3.1 24B Instruct is an upgraded variant of Mistral Small 3 (2501), featuring 24 billion parameters with
      advanced multimodal capabilities.
    match:
      equals: mistral-small-3.1-24b-instruct
    prices_checked: 2026-06-09
    price_comments: Imported from OpenRouter pricing; verify against Mistral pricing when native API pricing is published.
    prices:
      input_mtok: 0.351
      output_mtok: 0.555

  - id: mistral-small-3.2-24b-instruct
    name: Mistral Small 3.2 24B
    description: >-
      Mistral-Small-3.2-24B-Instruct-2506 is an updated 24B parameter model from Mistral optimized for instruction following,
      repetition reduction, and improved function calling.
    match:
      equals: mistral-small-3.2-24b-instruct
    prices_checked: 2026-06-09
    price_comments: Imported from OpenRouter pricing; verify against Mistral pricing when native API pricing is published.
    prices:
      input_mtok: 0.075
      output_mtok: 0.2

  - id: mistral-small-latest
    description: SOTA. Multimodal. Multilingual. Apache 2.0.
    name: Mistral Small 3.2
    match:
      equals: mistral-small-latest
    # No context_window: this alias pointed to the 128k mistral-small-2506 before repointing
    # to the 262,144 mistral-small-2603 — no single value is true for every request it matches
    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

  - id: voxtral-mini-2602
    name: Voxtral Mini Transcribe 2
    match:
      or:
        - equals: voxtral-mini-latest
        - equals: voxtral-mini-2602
    context_window: 16384
    prices_checked: 2026-10-14
    price_comments: See https://docs.mistral.ai/models/voxtral-mini-transcribe-26-02.
    prices:
      audio_hours: 0.18
      input_audio_hours: 0.18
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          audio_hours: 0.09
          input_audio_hours: 0.09

  - id: voxtral-small-24b-2507
    name: Voxtral Small 24B 2507
    description: >-
      Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining
      best-in-class text performance. It excels at speech transcription, translation and audio understanding.
    match:
      or:
        - equals: voxtral-small-24b-2507
        - equals: voxtral-small-2507
        - equals: voxtral-small-latest
    context_window: 32768
    prices_checked: 2026-10-14
    price_comments: >-
      Mistral raised output pricing from $0.30 to $0.40 per MTok on 2026-08-11.
      Ref: https://github.com/mistralai/platform-docs-public/commit/1996c3f1eca754d02436a37fcc899440794a18a5
    prices:
      - prices:
          input_mtok: 0.1
          cache_read_mtok: 0.01
          output_mtok: 0.3
      - constraint:
          start_date: 2026-08-11
        prices:
          input_mtok: 0.1
          cache_read_mtok: 0.01
          output_mtok: 0.4
    price_variants:
      - when: { processing: batch }
        constraint: { start_date: 2026-10-14 }
        prices:
          input_mtok: 0.05
          cache_read_mtok: 0.005
          output_mtok: 0.2