# Hand-owned model behavior overlay.
#
# Everything in this file is curated by hand and layered on top of API-sourced
# model data: oauth blocks, request patches, wire-name aliases, behavioral
# flags, and models that no public catalog provides.
#
# `models:` entries are self-complete. `rules` apply ONLY to API-sourced
# models, never to hand-owned `models:` entries. Legacy verbatim-preserved
# entries are grouped under a `# legacy -- pruning candidates` comment.

- provider: openai
  exclude:
    - "text-embedding-*"
    - "gpt-image-*"
    - "chatgpt-image-*"
    - "gpt-realtime*"
  rules:
    - match: "gpt-5.6-*"
      default_reasoning_effort: medium
    - match: "gpt-5.5"
      default_reasoning_effort: medium
    - match: "gpt-5.5-pro"
      default_reasoning_effort: high
    - match: "gpt-5.4"
      default_reasoning_effort: none
    - match: "gpt-5.4-pro"
      default_reasoning_effort: medium
    - match: "gpt-5.4-mini"
      default_reasoning_effort: none
    - match: "gpt-5.4-nano"
      default_reasoning_effort: none
    - match: "gpt-5.3-codex"
      default_reasoning_effort: medium
    - match: "gpt-5.2"
      default_reasoning_effort: none
    - match: "gpt-5.2-pro"
      default_reasoning_effort: medium
    - match: "gpt-5.1"
      default_reasoning_effort: none
    - match: "gpt-5.1-chat-latest"
      default_reasoning_effort: none
    - match: "gpt-5"
      default_reasoning_effort: medium
    - match: "gpt-5-chat-latest"
      default_reasoning_effort: medium
    - match: "o4-mini"
      system_prompt_prefix: Formatting re-enabled
      default_reasoning_effort: medium
      patch:
        body:
          max_tokens: null
          temperature: null
          top_p: null
    - match: "o3"
      system_prompt_prefix: Formatting re-enabled
      patch:
        body:
          max_tokens: null
          temperature: null
          top_p: null
    - match: "o3-mini"
      system_prompt_prefix: Formatting re-enabled
      patch:
        body:
          max_tokens: null
          temperature: null
          top_p: null
  models:
    - name: text-embedding-3-large
      type: embedding
      input_price: 0.13
      max_tokens_per_chunk: 8191
      default_chunk_size: 2000
      max_batch_size: 100
    - name: text-embedding-3-small
      type: embedding
      input_price: 0.02
      max_tokens_per_chunk: 8191
      default_chunk_size: 2000
      max_batch_size: 100
    # legacy -- pruning candidates
    - name: o4-mini-high
      max_output_tokens: 100000
      real_name: o4-mini
      max_input_tokens: 200000
      input_price: 1.1
      output_price: 4.4
      supports_vision: true
      supports_function_calling: true
      system_prompt_prefix: Formatting re-enabled
      patch:
        body:
          reasoning_effort: high
          max_tokens: null
          temperature: null
          top_p: null
    - name: o3-high
      real_name: o3
      max_input_tokens: 200000
      input_price: 2
      output_price: 8
      supports_vision: true
      supports_function_calling: true
      system_prompt_prefix: Formatting re-enabled
      patch:
        body:
          reasoning_effort: high
          max_tokens: null
          temperature: null
          top_p: null
    - name: o3-mini-high
      max_output_tokens: 100000
      real_name: o3-mini
      max_input_tokens: 200000
      input_price: 1.1
      output_price: 4.4
      supports_vision: true
      supports_function_calling: true
      system_prompt_prefix: Formatting re-enabled
      patch:
        body:
          reasoning_effort: high
          max_tokens: null
          temperature: null
          top_p: null
    - name: chat-latest
      max_input_tokens: 400000
      max_output_tokens: 128000
      input_price: 5
      output_price: 30
      supports_vision: true
      supports_function_calling: true
    - name: gpt-5.1-chat-latest
      max_input_tokens: 400000
      max_output_tokens: 128000
      input_price: 1.25
      output_price: 10
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [none, low, medium, high]
      default_reasoning_effort: none
    - name: gpt-5-chat-latest
      max_input_tokens: 400000
      max_output_tokens: 128000
      input_price: 1.25
      output_price: 10
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [minimal, low, medium, high]
      default_reasoning_effort: medium

- provider: gemini
  exclude:
    - "gemini-embedding-*"
    - "veo-*"
    - "*-tts"
    - "*-tts-preview"
    - "*-image-preview"
  rules:
    - match: "gemini-3.6-flash"
      default_reasoning_effort: medium
    - match: "gemini-3.5-flash"
      default_reasoning_effort: medium
    - match: "gemini-3-flash-preview"
      default_reasoning_effort: high
    - match: "gemini-3.1-flash-lite"
      default_reasoning_effort: minimal
    - match: "gemini-3.1-pro-preview"
      default_reasoning_effort: high
    - match: "gemini-3-pro-preview"
      default_reasoning_effort: high
    - match: "gemini-2.5-pro"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high]
  models:
    - name: text-embedding-004
      type: embedding
      input_price: 0
      max_tokens_per_chunk: 2048
      default_chunk_size: 1500
      max_batch_size: 100
    # legacy -- pruning candidates
    - name: gemini-3-pro-preview
      max_input_tokens: 1048576
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, high]
      default_reasoning_effort: high
    - name: gemma-3-27b-it
      max_input_tokens: 131072
      max_output_tokens: 8192
      input_price: 0
      output_price: 0

- provider: claude
  rules:
    - match: "claude-*"
      require_max_tokens: true
    - match: "claude-opus-5"
      default_reasoning_effort: high
    - match: "claude-fable-*"
      default_reasoning_effort: high
    - match: "claude-opus-4-8"
      default_reasoning_effort: high
    - match: "claude-opus-4-7"
      default_reasoning_effort: high
    - match: "claude-opus-4-6"
      default_reasoning_effort: high
    - match: "claude-sonnet-5"
      default_reasoning_effort: high
    - match: "claude-sonnet-4-6"
      default_reasoning_effort: high
  variants:
    - match: "claude-opus-4-6"
      suffix: ":thinking"
      real_name: claude-opus-4-6
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-sonnet-4-6"
      suffix: ":thinking"
      real_name: claude-sonnet-4-6
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-sonnet-4-5-20250929"
      suffix: ":thinking"
      real_name: claude-sonnet-4-5-20250929
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-haiku-4-5-20251001"
      suffix: ":thinking"
      real_name: claude-haiku-4-5-20251001
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-opus-4-5-20251101"
      suffix: ":thinking"
      real_name: claude-opus-4-5-20251101
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-opus-4-1-20250805"
      suffix: ":thinking"
      real_name: claude-opus-4-1-20250805
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-opus-4-20250514"
      suffix: ":thinking"
      real_name: claude-opus-4-20250514
      max_output_tokens: 24000
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
  models:
    # legacy -- pruning candidates
    - name: claude-opus-4-1-20250805
      max_input_tokens: 200000
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 15
      output_price: 75
      supports_vision: true
      supports_function_calling: true
    - name: claude-opus-4-20250514
      max_input_tokens: 200000
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 15
      output_price: 75
      supports_vision: true
      supports_function_calling: true

- provider: mistral
  exclude:
    - "voxtral-*"
  models:
    - name: mistral-embed
      type: embedding
      max_input_tokens: 8092
      input_price: 0.1
      max_tokens_per_chunk: 8092
      default_chunk_size: 2000
    # legacy -- pruning candidates
    - name: magistral-small-latest
      max_input_tokens: 131072
      input_price: 0.5
      output_price: 1.5
    - name: devstral-small-latest
      max_input_tokens: 262144
      input_price: 0.1
      output_price: 0.3
      supports_function_calling: true
    - name: ministral-14b-latest
      max_input_tokens: 262144
      input_price: 0.2
      output_price: 0.2
      supports_function_calling: true

- provider: ai21
  models:
    - name: jamba-large
      max_input_tokens: 256000
      input_price: 2
      output_price: 8
      supports_function_calling: true
    - name: jamba-mini
      max_input_tokens: 256000
      input_price: 0.2
      output_price: 0.4
      supports_function_calling: true

- provider: cohere
  models:
    - name: embed-v4.0
      type: embedding
      input_price: 0.12
      max_tokens_per_chunk: 2048
      default_chunk_size: 2000
      max_batch_size: 96
    - name: embed-english-v3.0
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    - name: embed-multilingual-v3.0
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    - name: rerank-v3.5
      type: reranker
      max_input_tokens: 4096
    - name: rerank-english-v3.0
      type: reranker
      max_input_tokens: 4096
    - name: rerank-multilingual-v3.0
      type: reranker
      max_input_tokens: 4096

- provider: xai
  exclude:
    - "grok-imagine-*"
  oauth:
    client_id: b1a00492-073a-47ea-816f-4c329264a828
    authorize_url: https://auth.x.ai/oauth2/authorize
    token_url: https://auth.x.ai/oauth2/token
    scopes:
      - openid
      - profile
      - email
      - offline_access
      - grok-cli:access
      - api:access
    redirect_port: 56121
    flow: pkce
    token_request_format: form_url_encoded
    extra_authorize_params:
      plan: generic
      referrer: coyote
    echo_pkce_in_token_exchange: true
  models:
    - name: grok-4.20
      real_name: grok-4.20-multi-agent-0309
      input_price: 1.25
      output_price: 2.5
      max_input_tokens: 1000000
      supports_function_calling: true
    - name: grok-4.20-reasoning
      real_name: grok-4.20-0309-reasoning
      input_price: 1.25
      output_price: 2.5
      max_input_tokens: 1000000
      supports_function_calling: true
    - name: grok-4.20-non-reasoning
      real_name: grok-4.20-0309-non-reasoning
      input_price: 1.25
      output_price: 2.5
      max_input_tokens: 1000000
      supports_function_calling: true
    - name: grok-4-1-fast-non-reasoning
      max_input_tokens: 1000000
      input_price: 0.2
      output_price: 0.5
      supports_function_calling: true
    - name: grok-4-1-fast-reasoning
      max_input_tokens: 1000000
      input_price: 0.2
      output_price: 0.5
      supports_function_calling: true
    - name: grok-code-fast-1
      max_input_tokens: 256000
      input_price: 0.2
      output_price: 1.5
      supports_function_calling: true
    - name: grok-3
      max_input_tokens: 131072
      input_price: 3
      output_price: 15
      supports_function_calling: true
    - name: grok-3-mini
      max_input_tokens: 131072
      input_price: 0.3
      output_price: 0.5
      supports_function_calling: true

- provider: perplexity
  models:
    # legacy -- pruning candidates
    - name: sonar-reasoning
      max_input_tokens: 128000
      input_price: 1
      output_price: 5

- provider: groq
  exclude:
    - "whisper-*"
  models:
    # legacy -- pruning candidates
    - name: meta-llama/llama-4-maverick-17b-128e-instruct
      max_input_tokens: 131072
      input_price: 0
      output_price: 0
      supports_vision: true
      supports_function_calling: true
    - name: meta-llama/llama-4-scout-17b-16e-instruct
      max_input_tokens: 131072
      input_price: 0
      output_price: 0
      supports_vision: true
      supports_function_calling: true
    - name: moonshotai/kimi-k2-instruct-0905
      max_input_tokens: 262144
      input_price: 0
      output_price: 0
      supports_function_calling: true
    - name: qwen/qwen3-32b
      max_input_tokens: 131072
      input_price: 0
      output_price: 0

- provider: vertexai
  exclude:
    - "gemini-embedding-*"
    - "*-tts"
  rules:
    - match: "claude-*"
      require_max_tokens: true
    - match: "gemini-3.5-flash"
      default_reasoning_effort: medium
    - match: "gemini-3-flash-preview"
      default_reasoning_effort: high
    - match: "gemini-3.1-flash-lite"
      default_reasoning_effort: minimal
    - match: "gemini-3.1-pro-preview"
      default_reasoning_effort: high
    - match: "gemini-3-pro-preview"
      default_reasoning_effort: high
    - match: "gemini-2.5-flash"
      default_reasoning_effort: medium
      reasoning_levels: [low, medium, high]
    - match: "gemini-2.5-pro"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high]
    - match: "claude-opus-5"
      default_reasoning_effort: high
    - match: "claude-fable-*"
      default_reasoning_effort: high
    - match: "claude-opus-4-8"
      default_reasoning_effort: high
    - match: "claude-opus-4-7"
      default_reasoning_effort: high
    - match: "claude-sonnet-5"
      default_reasoning_effort: high
    - match: "claude-sonnet-4-6"
      default_reasoning_effort: high
  variants:
    - match: "claude-opus-4-6"
      suffix: ":thinking"
      real_name: claude-opus-4-6
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-sonnet-4-6"
      suffix: ":thinking"
      real_name: claude-sonnet-4-6
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-opus-4-5@20251101"
      suffix: ":thinking"
      real_name: claude-opus-4-5@20251101
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-sonnet-4-5@20250929"
      suffix: ":thinking"
      real_name: claude-sonnet-4-5@20250929
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
    - match: "claude-haiku-4-5@20251001"
      suffix: ":thinking"
      real_name: claude-haiku-4-5@20251001
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          temperature: null
          top_p: null
          thinking:
            type: enabled
            budget_tokens: 16000
  models:
    - name: text-embedding-005
      type: embedding
      max_input_tokens: 20000
      input_price: 0.025
      max_tokens_per_chunk: 2048
      default_chunk_size: 1500
      max_batch_size: 5
    - name: text-multilingual-embedding-002
      type: embedding
      max_input_tokens: 20000
      input_price: 0.2
      max_tokens_per_chunk: 2048
      default_chunk_size: 1500
      max_batch_size: 5
    # legacy -- pruning candidates
    - name: gemini-3-pro-preview
      max_input_tokens: 1048576
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, high]
      default_reasoning_effort: high
    - name: claude-opus-5
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_function_calling: true
      supports_vision: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-fable-5-1
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 10
      output_price: 50
      supports_function_calling: true
      supports_vision: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-fable-5
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 10
      output_price: 50
      supports_function_calling: true
      supports_vision: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-opus-4-8
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-opus-4-7
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-opus-4-6
      max_input_tokens: 200000
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_vision: true
      supports_function_calling: true
    - name: claude-sonnet-5
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 3
      output_price: 15
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: claude-sonnet-4-6
      max_input_tokens: 200000
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 3
      output_price: 15
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, max]
      default_reasoning_effort: high
    - name: claude-3-5-haiku@20241022
      max_input_tokens: 200000
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 0.8
      output_price: 4
      supports_vision: true
      supports_function_calling: true
    - name: mistral-small-2503
      max_input_tokens: 32000
      input_price: 0.1
      output_price: 0.3
      supports_function_calling: true
    - name: codestral-2501
      max_input_tokens: 256000
      input_price: 0.3
      output_price: 0.9
      supports_function_calling: true

- provider: bedrock
  rules:
    - match: "us.anthropic.claude-*"
      require_max_tokens: true
    - match: "us.meta.llama*"
      require_max_tokens: true
    - match: "us.anthropic.claude-opus-5"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-fable-*"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-opus-4-8"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-opus-4-7"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-opus-4-6-v1"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-sonnet-5"
      default_reasoning_effort: high
    - match: "us.anthropic.claude-sonnet-4-6"
      default_reasoning_effort: high
  variants:
    - match: "us.anthropic.claude-opus-4-6-v1"
      suffix: ":thinking"
      real_name: us.anthropic.claude-opus-4-6-v1
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          inferenceConfig:
            temperature: null
            topP: null
          additionalModelRequestFields:
            thinking:
              type: enabled
              budget_tokens: 16000
    - match: "us.anthropic.claude-sonnet-4-6"
      suffix: ":thinking"
      real_name: us.anthropic.claude-sonnet-4-6
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          inferenceConfig:
            temperature: null
            topP: null
          additionalModelRequestFields:
            thinking:
              type: enabled
              budget_tokens: 16000
    - match: "us.anthropic.claude-opus-4-5-20251101-v1:0"
      suffix: ":thinking"
      real_name: us.anthropic.claude-opus-4-5-20251101-v1:0
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          inferenceConfig:
            temperature: null
            topP: null
          additionalModelRequestFields:
            thinking:
              type: enabled
              budget_tokens: 16000
    - match: "us.anthropic.claude-sonnet-4-5-20250929-v1:0"
      suffix: ":thinking"
      real_name: us.anthropic.claude-sonnet-4-5-20250929-v1:0
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          inferenceConfig:
            temperature: null
            topP: null
          additionalModelRequestFields:
            thinking:
              type: enabled
              budget_tokens: 16000
    - match: "us.anthropic.claude-haiku-4-5-20251001-v1:0"
      suffix: ":thinking"
      real_name: us.anthropic.claude-haiku-4-5-20251001-v1:0
      max_output_tokens: 24000
      supports_function_calling: false
      reasoning_levels: []
      patch:
        body:
          inferenceConfig:
            temperature: null
            topP: null
          additionalModelRequestFields:
            thinking:
              type: enabled
              budget_tokens: 16000
  models:
    - name: cohere.embed-english-v3
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    - name: cohere.embed-multilingual-v3
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    # legacy -- pruning candidates
    - name: us.meta.llama3-3-70b-instruct-v1:0
      max_input_tokens: 131072
      max_output_tokens: 8192
      require_max_tokens: true
      input_price: 0.72
      output_price: 0.72
      supports_function_calling: true
    - name: us.amazon.nova-premier-v1:0
      max_input_tokens: 300000
      max_output_tokens: 5120
      input_price: 2.5
      output_price: 12.5
    - name: us.amazon.nova-pro-v1:0
      max_input_tokens: 300000
      max_output_tokens: 5120
      input_price: 0.8
      output_price: 3.2
      supports_vision: true
    - name: us.amazon.nova-lite-v1:0
      max_input_tokens: 300000
      max_output_tokens: 5120
      input_price: 0.06
      output_price: 0.24
      supports_vision: true
    - name: us.amazon.nova-micro-v1:0
      max_input_tokens: 128000
      max_output_tokens: 5120
      input_price: 0.035
      output_price: 0.14

- provider: cloudflare
  rules:
    - match: "@cf/*"
      require_max_tokens: true
  models:
    - name: '@cf/baai/bge-large-en-v1.5'
      type: embedding
      input_price: 0
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 100
    # legacy -- pruning candidates
    - name: '@cf/google/gemma-3-12b-it'
      max_input_tokens: 131072
      max_output_tokens: 2048
      require_max_tokens: true
      input_price: 0
      output_price: 0

- provider: ernie
  models:
    - name: ernie-4.5-turbo-128k
      max_input_tokens: 131072
      input_price: 0.112
      output_price: 0.448
    - name: ernie-4.5-turbo-vl-32k
      max_input_tokens: 32768
      input_price: 0.42
      output_price: 1.26
      supports_vision: true
    - name: ernie-5.0-thinking-preview
      max_input_tokens: 131072
      input_price: 1.4
      output_price: 5.6
    - name: ernie-x1.1-preview
      max_input_tokens: 65536
      input_price: 0.14
      output_price: 0.56
    - name: bge-large-zh
      type: embedding
      input_price: 0.07
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 16
    - name: bge-large-en
      type: embedding
      input_price: 0.07
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 16
    - name: bce-reranker-base
      type: reranker
      max_input_tokens: 1024
      input_price: 0.07

- provider: qianwen
  exclude:
    - "*-realtime"
  rules:
    - match: "qwen3.5-plus"
      patch:
        body:
          enable_thinking: false
    - match: "qwen3.5-397b-a17b"
      patch:
        body:
          enable_thinking: false
  variants:
    - match: "qwen3-max"
      suffix: ":thinking"
      real_name: qwen3-max
      patch:
        body:
          enable_thinking: true
  models:
    - name: qwen3.5-plus:thinking
      real_name: qwen3.5-plus
      max_input_tokens: 262144
      supports_function_calling: true
    - name: qwen3.5-397b-a17b:thinking
      real_name: qwen3.5-397b-a17b
      max_input_tokens: 262144
      supports_function_calling: true
    - name: text-embedding-v4
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 8192
      default_chunk_size: 2000
      max_batch_size: 10
    - name: text-embedding-v3
      type: embedding
      input_price: 0.1
      max_tokens_per_chunk: 8192
      default_chunk_size: 2000
      max_batch_size: 10
    # legacy -- pruning candidates
    - name: qwen3-vl-flash
      max_input_tokens: 262144
      supports_vision: true
    - name: qwen3-235b-a22b-instruct-2507
      max_input_tokens: 131072
      input_price: 0.28
      output_price: 1.12
      supports_function_calling: true
    - name: qwen3-235b-a22b-thinking-2507
      max_input_tokens: 131072
      input_price: 0.28
      output_price: 2.8
    - name: qwen3-30b-a3b-instruct-2507
      max_input_tokens: 131072
      input_price: 0.105
      output_price: 0.42
      supports_function_calling: true
    - name: qwen3-30b-a3b-thinking-2507
      max_input_tokens: 131072
      input_price: 0.105
      output_price: 1.05
    - name: qwen3-vl-32b-instruct
      max_input_tokens: 131072
      input_price: 0.28
      output_price: 1.12
      supports_vision: true
    - name: qwen3-vl-8b-instruct
      max_input_tokens: 131072
      input_price: 0.07
      output_price: 0.28
      supports_vision: true
    - name: qwen3-coder-next
      max_input_tokens: 262144

- provider: hunyuan
  models:
    - name: hunyuan-2.0-instruct-20251111
      max_input_tokens: 131072
      input_price: 0.112
      output_price: 0.28
      supports_function_calling: true
    - name: hunyuan-2.0-thinking-20251109
      max_input_tokens: 131072
      input_price: 0.14
      output_price: 0.56
      supports_function_calling: true
    - name: hunyuan-vision-1.5-instruct
      max_input_tokens: 24576
      input_price: 0.42
      output_price: 1.26
      supports_vision: true
    - name: hunyuan-embedding
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 1024
      default_chunk_size: 1000
      max_batch_size: 100

- provider: deepseek
  models:
    # legacy -- pruning candidates
    - name: deepseek-chat
      max_input_tokens: 64000
      max_output_tokens: 8192
      input_price: 0.56
      output_price: 1.68
      supports_function_calling: true
    - name: deepseek-reasoner
      max_input_tokens: 64000
      max_output_tokens: 32768
      input_price: 0.56
      output_price: 1.68

- provider: zhipuai
  variants:
    - match: "glm-5"
      suffix: ":instruct"
      real_name: glm-5
      patch:
        body:
          thinking:
            type: disabled
    - match: "glm-4.7"
      suffix: ":instruct"
      real_name: glm-4.7
      patch:
        body:
          thinking:
            type: disabled
  models:
    - name: embedding-3
      type: embedding
      max_input_tokens: 8192
      input_price: 0.07
      max_tokens_per_chunk: 8192
      default_chunk_size: 2000
    - name: rerank
      type: reranker
      max_input_tokens: 4096
      input_price: 0.112
    # legacy -- pruning candidates
    - name: glm-4.6v-flash
      max_input_tokens: 65536
      input_price: 0
      output_price: 0
      supports_vision: true

- provider: minimax
  models:
    # legacy -- pruning candidates
    - name: minimax-m3
      max_input_tokens: 1000000
      input_price: 4.2
      output_price: 16.8
      supports_function_calling: true
    - name: minimax-m2.7
      max_input_tokens: 204800
      input_price: 0.294
      output_price: 1.176
      supports_function_calling: true
    - name: minimax-m2.5
      max_input_tokens: 204800
      input_price: 0.294
      output_price: 1.176
      supports_function_calling: true
    - name: minimax-m2.5-highspeed
      max_input_tokens: 204800
      input_price: 0.588
      output_price: 2.352
      supports_function_calling: true
    - name: minimax-m2.1
      max_input_tokens: 204800
      input_price: 0.294
      output_price: 1.176
      supports_function_calling: true
    - name: minimax-m2.1-highspeed
      max_input_tokens: 204800
      input_price: 0.588
      output_price: 2.352
      supports_function_calling: true

- provider: openrouter
  rules:
    - match: "anthropic/claude-*"
      require_max_tokens: true
    - match: "openai/gpt-5.6-*"
      default_reasoning_effort: medium
      reasoning_levels: [none, low, medium, high, xhigh, max]
    - match: "openai/gpt-5.5"
      default_reasoning_effort: medium
      reasoning_levels: [none, low, medium, high, xhigh]
    - match: "openai/gpt-5.5-pro"
      default_reasoning_effort: high
      reasoning_levels: [medium, high, xhigh]
    - match: "openai/gpt-5.4"
      default_reasoning_effort: none
      reasoning_levels: [none, low, medium, high, xhigh]
    - match: "openai/gpt-5.4-pro"
      default_reasoning_effort: medium
      reasoning_levels: [medium, high, xhigh]
    - match: "openai/gpt-5.4-mini"
      default_reasoning_effort: none
      reasoning_levels: [none, low, medium, high, xhigh]
    - match: "openai/gpt-5.4-nano"
      default_reasoning_effort: none
      reasoning_levels: [none, low, medium, high, xhigh]
    - match: "openai/gpt-5.3-codex"
      default_reasoning_effort: medium
      reasoning_levels: [low, medium, high, xhigh]
    - match: "openai/gpt-5.2"
      default_reasoning_effort: none
      reasoning_levels: [none, low, medium, high, xhigh]
    - match: "openai/gpt-5.2-pro"
      default_reasoning_effort: medium
      reasoning_levels: [medium, high, xhigh]
    - match: "openai/gpt-5"
      default_reasoning_effort: medium
      reasoning_levels: [minimal, low, medium, high]
    - match: "google/gemini-3.5-flash"
      default_reasoning_effort: medium
      reasoning_levels: [minimal, low, medium, high]
    - match: "google/gemini-3-flash-preview"
      default_reasoning_effort: high
      reasoning_levels: [minimal, low, medium, high]
    - match: "google/gemini-3.1-flash-lite"
      default_reasoning_effort: minimal
      reasoning_levels: [minimal, low, medium, high]
    - match: "google/gemini-3.1-pro-preview"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high]
    - match: "google/gemini-3-pro-preview"
      default_reasoning_effort: high
    - match: "google/gemini-2.5-flash"
      default_reasoning_effort: low
      reasoning_levels: [low, medium, high]
    - match: "google/gemini-2.5-pro"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high]
    - match: "anthropic/claude-opus-5"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high, xhigh, max]
    - match: "anthropic/claude-fable-*"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high, xhigh, max]
    - match: "anthropic/claude-opus-4-8"
      default_reasoning_effort: high
    - match: "anthropic/claude-opus-4-7"
      default_reasoning_effort: high
    - match: "anthropic/claude-opus-4.6"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high, max]
    - match: "anthropic/claude-sonnet-5"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high, xhigh, max]
    - match: "anthropic/claude-sonnet-4.6"
      default_reasoning_effort: high
      reasoning_levels: [low, medium, high, max]
    - match: "perplexity/sonar-reasoning-pro"
      patch:
        body:
          include_reasoning: true
    - match: "perplexity/sonar-deep-research"
      patch:
        body:
          include_reasoning: true
  models:
    # legacy -- pruning candidates
    - name: google/gemini-3-pro-preview
      max_input_tokens: 1048576
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, high]
      default_reasoning_effort: high
    - name: google/gemini-2.0-flash-001
      max_input_tokens: 1000000
      input_price: 0.15
      output_price: 0.6
      supports_vision: true
      supports_function_calling: true
    - name: google/gemini-2.0-flash-lite-001
      max_input_tokens: 1048576
      input_price: 0.075
      output_price: 0.3
      supports_vision: true
      supports_function_calling: true
    - name: anthropic/claude-fable-5-1
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 10
      output_price: 50
      supports_function_calling: true
      supports_vision: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: anthropic/claude-opus-4-8
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: anthropic/claude-opus-4-7
      max_input_tokens: 1000000
      max_output_tokens: 128000
      require_max_tokens: true
      input_price: 5
      output_price: 25
      supports_vision: true
      supports_function_calling: true
      reasoning_levels: [low, medium, high, xhigh, max]
      default_reasoning_effort: high
    - name: mistralai/devstral-small
      max_input_tokens: 131072
      input_price: 0.07
      output_price: 0.28
      supports_function_calling: true
    - name: ai21/jamba-large-1.7
      max_input_tokens: 256000
      input_price: 2
      output_price: 8
      supports_function_calling: true
    - name: x-ai/grok-4.1-fast
      max_input_tokens: 2000000
      input_price: 0.2
      output_price: 0.5
      supports_function_calling: true
    - name: x-ai/grok-code-fast-1
      max_input_tokens: 256000
      input_price: 0.2
      output_price: 1.5
      supports_function_calling: true

# Fully hand-owned: upstream catalogs track GitHub Copilot ids, which do not
# match the GitHub Models API ids coyote uses.
- provider: github
  models:
    - name: gpt-5
      max_input_tokens: 400000
      max_output_tokens: 128000
      supports_vision: true
      supports_function_calling: true
    - name: gpt-5-mini
      max_input_tokens: 400000
      max_output_tokens: 128000
      supports_vision: true
      supports_function_calling: true
    - name: gpt-5-nano
      max_input_tokens: 400000
      max_output_tokens: 128000
      supports_vision: true
      supports_function_calling: true
    - name: gpt-4.1
      max_input_tokens: 1047576
      max_output_tokens: 32768
      supports_vision: true
      supports_function_calling: true
    - name: gpt-4o
      max_input_tokens: 128000
      max_output_tokens: 16384
      supports_function_calling: true
    - name: text-embedding-3-large
      type: embedding
      max_tokens_per_chunk: 8191
      default_chunk_size: 2000
      max_batch_size: 100
    - name: text-embedding-3-small
      type: embedding
      max_tokens_per_chunk: 8191
      default_chunk_size: 2000
      max_batch_size: 100
    - name: llama-4-maverick-17b-128e-instruct-fp8
      max_input_tokens: 1048576
      supports_vision: true
    - name: llama-4-scout-17b-16e-instruct
      max_input_tokens: 327680
      supports_vision: true
    - name: llama-3.3-70b-instruct
      max_input_tokens: 131072
    - name: mistral-medium-2505
      max_input_tokens: 131072
      supports_function_calling: true
    - name: mistral-small-2503
      max_input_tokens: 131072
      supports_function_calling: true
    - name: codestral-2501
      max_input_tokens: 256000
      supports_function_calling: true
    - name: cohere-embed-v3-english
      type: embedding
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    - name: cohere-embed-v3-multilingual
      type: embedding
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 96
    - name: deepseek-r1-0528
      max_input_tokens: 163840
    - name: deepseek-v3-0324
      max_input_tokens: 163840
    - name: mai-ds-r1
      max_input_tokens: 163840
    - name: phi-4
      max_input_tokens: 16384
    - name: phi-4-mini-instruct
      max_input_tokens: 131072
    - name: phi-4-reasoning
      max_input_tokens: 33792
    - name: phi-4-mini-reasoning
      max_input_tokens: 131072
    - name: grok-3
      max_input_tokens: 131072
    - name: grok-3-mini
      max_input_tokens: 131072

- provider: deepinfra
  models:
    - name: BAAI/bge-large-en-v1.5
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 100
    - name: BAAI/bge-m3
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 8192
      default_chunk_size: 2000
      max_batch_size: 100
    - name: intfloat/e5-large-v2
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 100
    - name: intfloat/multilingual-e5-large
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 100
    - name: thenlper/gte-large
      type: embedding
      input_price: 0.01
      max_tokens_per_chunk: 512
      default_chunk_size: 1000
      max_batch_size: 100
    # legacy -- pruning candidates
    - name: Qwen/Qwen3-Max-Thinking
      max_input_tokens: 262144
      input_price: 1.2
      output_price: 6
      supports_function_calling: true
    - name: Qwen/Qwen3-Next-80B-A3B-Thinking
      max_input_tokens: 262144
      input_price: 0.14
      output_price: 1.4
    - name: Qwen/Qwen3-235B-A22B-Thinking-2507
      max_input_tokens: 131072
      input_price: 0.13
      output_price: 0.6
    - name: Qwen/Qwen3-Coder-480B-A35B-Instruct
      max_input_tokens: 131072
      input_price: 0.4
      output_price: 1.6
      supports_function_calling: true
    - name: Qwen/Qwen3-Coder-30B-A3B-Instruct
      max_input_tokens: 262144
      input_price: 0.07
      output_price: 0.27
      supports_function_calling: true
    - name: Qwen/Qwen3-VL-8B-Instruct
      max_input_tokens: 262144
      input_price: 0.18
      output_price: 0.69
      supports_vision: true
    - name: google/gemma-3-27b-it
      max_input_tokens: 131072
      input_price: 0.1
      output_price: 0.2
    - name: mistralai/Mistral-Small-3.2-24B-Instruct-2506
      max_input_tokens: 32768
      input_price: 0.06
      output_price: 0.12
    - name: moonshotai/Kimi-K2-Instruct-0905
      max_input_tokens: 262144
      input_price: 0.5
      output_price: 2.0
      supports_function_calling: true
    - name: moonshotai/Kimi-K2-Thinking
      max_input_tokens: 262144
      input_price: 0.55
      output_price: 2.5
      supports_function_calling: true
    - name: MiniMaxAI/MiniMax-M2.1
      max_input_tokens: 196608
      input_price: 0.27
      output_price: 0.95
      supports_function_calling: true
    - name: zai-org/GLM-4.6V
      max_input_tokens: 131072
      input_price: 0.3
      output_price: 0.9
      supports_vision: true

- provider: jina
  models:
    - name: jina-embeddings-v3
      type: embedding
      input_price: 0
      max_tokens_per_chunk: 8192
      default_chunk_size: 2000
      max_batch_size: 100
    - name: jina-clip-v2
      type: embedding
      input_price: 0
      max_tokens_per_chunk: 8192
      default_chunk_size: 1500
      max_batch_size: 100
    - name: jina-colbert-v2
      type: embedding
      input_price: 0
      max_tokens_per_chunk: 8192
      default_chunk_size: 1500
      max_batch_size: 100
    - name: jina-reranker-v2-base-multilingual
      type: reranker
      max_input_tokens: 8192
      input_price: 0
    - name: jina-colbert-v2
      type: reranker
      max_input_tokens: 8192
      input_price: 0

- provider: voyageai
  models:
    - name: voyage-3-large
      type: embedding
      max_input_tokens: 120000
      input_price: 0.18
      max_tokens_per_chunk: 32000
      default_chunk_size: 2000
      max_batch_size: 128
    - name: voyage-3
      type: embedding
      max_input_tokens: 320000
      input_price: 0.06
      max_tokens_per_chunk: 32000
      default_chunk_size: 2000
      max_batch_size: 128
    - name: voyage-3-lite
      type: embedding
      max_input_tokens: 1000000
      input_price: 0.02
      max_tokens_per_chunk: 32000
      default_chunk_size: 1000
      max_batch_size: 128
    - name: rerank-2
      type: reranker
      max_input_tokens: 16000
      input_price: 0.05
    - name: rerank-2-lite
      type: reranker
      max_input_tokens: 8000
      input_price: 0.02