omni-dev 0.24.0

# AI Model Specifications
# This file defines the capabilities and limits for all supported AI models
# Used by omni-dev to configure API requests correctly
# Last updated: February 2026

models:
  # Claude 4.6 Series (Current Generation)
  - provider: "claude"
    model: "Claude Opus 4.6"
    api_identifier: "claude-opus-4-6"
    max_output_tokens: 128000
    input_context: 200000
    generation: 4.6
    tier: "flagship"
    beta_headers:
      - key: "anthropic-beta"
        value: "context-1m-2025-08-07"
        input_context: 1000000

  - provider: "claude"
    model: "Claude Sonnet 4.6"
    api_identifier: "claude-sonnet-4-6"
    max_output_tokens: 64000
    input_context: 200000
    generation: 4.6
    tier: "balanced"
    beta_headers:
      - key: "anthropic-beta"
        value: "context-1m-2025-08-07"
        input_context: 1000000

  # Claude 4.5 Series (Current Generation)
  - provider: "claude"
    model: "Claude Sonnet 4.5"
    api_identifier: "claude-sonnet-4-5-20250929"
    max_output_tokens: 64000
    input_context: 200000
    generation: 4.5
    tier: "balanced"
    beta_headers:
      - key: "anthropic-beta"
        value: "context-1m-2025-08-07"
        input_context: 1000000

  - provider: "claude"
    model: "Claude Haiku 4.5"
    api_identifier: "claude-haiku-4-5-20251001"
    max_output_tokens: 64000
    input_context: 200000
    generation: 4.5
    tier: "fast"

  - provider: "claude"
    model: "Claude Opus 4.5"
    api_identifier: "claude-opus-4-5-20251101"
    max_output_tokens: 64000
    input_context: 200000
    generation: 4.5
    tier: "flagship"
    legacy: true

  # Claude 4.x Series (Legacy)
  - provider: "claude"
    model: "Claude Opus 4.1"
    api_identifier: "claude-opus-4-1-20250805"
    max_output_tokens: 32000
    input_context: 200000
    generation: 4.1
    tier: "flagship"
    legacy: true

  - provider: "claude"
    model: "Claude Opus 4"
    api_identifier: "claude-opus-4-20250514"
    max_output_tokens: 32000
    input_context: 200000
    generation: 4
    tier: "flagship"
    legacy: true

  - provider: "claude"
    model: "Claude Sonnet 4"
    api_identifier: "claude-sonnet-4-20250514"
    max_output_tokens: 64000
    input_context: 200000
    generation: 4
    tier: "balanced"
    legacy: true
    beta_headers:
      - key: "anthropic-beta"
        value: "context-1m-2025-08-07"
        input_context: 1000000

  # Claude 3.x Series (Legacy - deprecated, retiring Feb 19, 2026)
  - provider: "claude"
    model: "Claude Sonnet 3.7"
    api_identifier: "claude-3-7-sonnet-20250219"
    max_output_tokens: 64000
    input_context: 200000
    generation: 3.7
    tier: "balanced"
    legacy: true
    beta_headers:
      - key: "anthropic-beta"
        value: "output-128k-2025-02-19"
        max_output_tokens: 128000
      - key: "anthropic-beta"
        value: "context-1m-2025-08-07"
        input_context: 1000000

  - provider: "claude"
    model: "Claude Haiku 3.5"
    api_identifier: "claude-3-5-haiku-20241022"
    max_output_tokens: 8192
    input_context: 200000
    generation: 3.5
    tier: "fast"
    legacy: true

  - provider: "claude"
    model: "Claude Haiku 3"
    api_identifier: "claude-3-haiku-20240307"
    max_output_tokens: 4096
    input_context: 200000
    generation: 3
    tier: "fast"
    legacy: true

  # Claude 3 Series (Retired - kept for Bedrock/AWS fuzzy matching)
  - provider: "claude"
    model: "Claude Opus 3"
    api_identifier: "claude-3-opus-20240229"
    max_output_tokens: 4096
    input_context: 200000
    generation: 3
    tier: "flagship"
    legacy: true

  - provider: "claude"
    model: "Claude Sonnet 3.5"
    api_identifier: "claude-3-5-sonnet-20241022"
    max_output_tokens: 8192
    input_context: 200000
    generation: 3.5
    tier: "balanced"
    legacy: true

  # OpenAI GPT-5 Series (Current Generation)
  - provider: "openai"
    model: "GPT-5.2"
    api_identifier: "gpt-5.2"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5.2
    tier: "flagship"

  - provider: "openai"
    model: "GPT-5.2 Pro"
    api_identifier: "gpt-5.2-pro"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5.2
    tier: "flagship"

  - provider: "openai"
    model: "GPT-5"
    api_identifier: "gpt-5"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5
    tier: "flagship"

  - provider: "openai"
    model: "GPT-5 Pro"
    api_identifier: "gpt-5-pro"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5
    tier: "flagship"

  - provider: "openai"
    model: "GPT-5 Mini"
    api_identifier: "gpt-5-mini"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5
    tier: "balanced"

  - provider: "openai"
    model: "GPT-5 Nano"
    api_identifier: "gpt-5-nano"
    max_output_tokens: 128000
    input_context: 400000
    generation: 5
    tier: "fast"

  # OpenAI o-Series Reasoning Models
  - provider: "openai"
    model: "o3"
    api_identifier: "o3"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "flagship"

  - provider: "openai"
    model: "o3-pro"
    api_identifier: "o3-pro"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "flagship"

  - provider: "openai"
    model: "o3-mini"
    api_identifier: "o3-mini"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "fast"

  - provider: "openai"
    model: "o4-mini"
    api_identifier: "o4-mini"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "fast"
    legacy: true

  - provider: "openai"
    model: "o1"
    api_identifier: "o1"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "flagship"
    legacy: true

  - provider: "openai"
    model: "o1-pro"
    api_identifier: "o1-pro"
    max_output_tokens: 100000
    input_context: 200000
    generation: 5
    tier: "flagship"
    legacy: true

  - provider: "openai"
    model: "o1-mini"
    api_identifier: "o1-mini"
    max_output_tokens: 65536
    input_context: 128000
    generation: 5
    tier: "fast"
    legacy: true

  - provider: "openai"
    model: "o1-preview"
    api_identifier: "o1-preview"
    max_output_tokens: 32768
    input_context: 128000
    generation: 5
    tier: "flagship"
    legacy: true

  # OpenAI GPT-4.1 Series (Legacy - retiring Feb 2026)
  - provider: "openai"
    model: "GPT-4.1"
    api_identifier: "gpt-4.1"
    max_output_tokens: 32768
    input_context: 1047576
    generation: 4.1
    tier: "flagship"
    legacy: true

  - provider: "openai"
    model: "GPT-4.1 Mini"
    api_identifier: "gpt-4.1-mini"
    max_output_tokens: 32768
    input_context: 1047576
    generation: 4.1
    tier: "balanced"
    legacy: true

  - provider: "openai"
    model: "GPT-4.1 Nano"
    api_identifier: "gpt-4.1-nano"
    max_output_tokens: 32768
    input_context: 1047576
    generation: 4.1
    tier: "fast"
    legacy: true

  # OpenAI GPT-4o Series (Legacy - deprecated)
  - provider: "openai"
    model: "GPT-4o"
    api_identifier: "gpt-4o"
    max_output_tokens: 16384
    input_context: 128000
    generation: 4
    tier: "flagship"
    legacy: true

  - provider: "openai"
    model: "GPT-4o Mini"
    api_identifier: "gpt-4o-mini"
    max_output_tokens: 16384
    input_context: 128000
    generation: 4
    tier: "fast"
    legacy: true

  # OpenAI GPT-4 Turbo (Legacy - deprecated)
  - provider: "openai"
    model: "GPT-4 Turbo"
    api_identifier: "gpt-4-turbo"
    max_output_tokens: 4096
    input_context: 128000
    generation: 4
    tier: "balanced"
    legacy: true

  # Google Gemini 3 Series (Preview - Latest)
  - provider: "gemini"
    model: "Gemini 3 Pro"
    api_identifier: "gemini-3-pro-preview"
    max_output_tokens: 65536
    input_context: 1048576
    generation: 3.0
    tier: "flagship"

  - provider: "gemini"
    model: "Gemini 3 Flash"
    api_identifier: "gemini-3-flash-preview"
    max_output_tokens: 65536
    input_context: 1048576
    generation: 3.0
    tier: "balanced"

  # Google Gemini 2.5 Series (Stable)
  - provider: "gemini"
    model: "Gemini 2.5 Pro"
    api_identifier: "gemini-2.5-pro"
    max_output_tokens: 65536
    input_context: 1048576
    generation: 2.5
    tier: "flagship"

  - provider: "gemini"
    model: "Gemini 2.5 Flash"
    api_identifier: "gemini-2.5-flash"
    max_output_tokens: 65536
    input_context: 1048576
    generation: 2.5
    tier: "balanced"

  - provider: "gemini"
    model: "Gemini 2.5 Flash-Lite"
    api_identifier: "gemini-2.5-flash-lite"
    max_output_tokens: 65536
    input_context: 1048576
    generation: 2.5
    tier: "fast"

  # Google Gemini 2.0 Series (Legacy - deprecating Mar 2026)
  - provider: "gemini"
    model: "Gemini 2.0 Flash"
    api_identifier: "gemini-2.0-flash"
    max_output_tokens: 8192
    input_context: 1048576
    generation: 2.0
    tier: "fast"
    legacy: true

  - provider: "gemini"
    model: "Gemini 2.0 Flash-Lite"
    api_identifier: "gemini-2.0-flash-lite"
    max_output_tokens: 8192
    input_context: 1048576
    generation: 2.0
    tier: "fast"
    legacy: true

  # Google Gemini 1.5 Series (Retired - kept for reference)
  - provider: "gemini"
    model: "Gemini 1.5 Pro"
    api_identifier: "gemini-1.5-pro"
    max_output_tokens: 8192
    input_context: 2097152
    generation: 1.5
    tier: "flagship"
    legacy: true

  - provider: "gemini"
    model: "Gemini 1.5 Flash"
    api_identifier: "gemini-1.5-flash"
    max_output_tokens: 8192
    input_context: 1048576
    generation: 1.5
    tier: "balanced"
    legacy: true

# Provider-specific configurations
providers:
  claude:
    name: "Anthropic Claude"
    api_base: "https://api.anthropic.com/v1"
    default_model: "claude-sonnet-4-6"

    # Model tier descriptions for Claude
    tiers:
      flagship:
        description: "Most capable models for complex reasoning and analysis"
        use_cases: ["complex analysis", "research", "advanced reasoning"]

      balanced:
        description: "Good balance of capability and speed"
        use_cases: ["general tasks", "coding", "writing"]

      fast:
        description: "Optimized for speed and cost efficiency"
        use_cases: ["simple tasks", "high-volume processing", "quick responses"]

    # Default fallback limits for unknown Claude models
    defaults:
      max_output_tokens: 4096
      input_context: 200000

  openai:
    name: "OpenAI"
    api_base: "https://api.openai.com/v1"
    default_model: "gpt-5-mini"

    # Model tier descriptions for OpenAI
    tiers:
      flagship:
        description: "Most capable models for complex reasoning and analysis"
        use_cases: ["complex analysis", "research", "advanced reasoning"]

      balanced:
        description: "Good balance of capability and speed"
        use_cases: ["general tasks", "coding", "writing"]

      fast:
        description: "Optimized for speed and cost efficiency"
        use_cases: ["simple tasks", "high-volume processing", "quick responses"]

    # Default fallback limits for unknown OpenAI models
    defaults:
      max_output_tokens: 16384
      input_context: 128000

  gemini:
    name: "Google Gemini"
    api_base: "https://generativelanguage.googleapis.com/v1beta"
    default_model: "gemini-2.5-flash"

    # Model tier descriptions for Gemini
    tiers:
      flagship:
        description: "Most capable models for complex reasoning and analysis"
        use_cases: ["complex analysis", "research", "advanced reasoning"]

      balanced:
        description: "Good balance of capability and speed"
        use_cases: ["general tasks", "coding", "writing"]

      fast:
        description: "Optimized for speed and cost efficiency"
        use_cases: ["simple tasks", "high-volume processing", "quick responses"]

    # Default fallback limits for unknown Gemini models
    defaults:
      max_output_tokens: 8192
      input_context: 1048576