models.py

# models.py

from dataclasses import dataclass
from typing import List, Optional

@dataclass
class ModelInfo:
    """
    Represents metadata for an inference model.

    Attributes:
        name: Human‑readable name of the model.
        id: Unique model identifier (HF or routed provider).
        description: Short description of the model's capabilities.
        default_provider: Preferred inference provider
                          ("auto", "groq", "openai", "gemini", "fireworks").
    """
    name: str
    id: str
    description: str
    default_provider: str = "auto"


# Registry of supported models
AVAILABLE_MODELS: List[ModelInfo] = [
    ModelInfo(
        name="Moonshot Kimi‑K2",
        id="moonshotai/Kimi‑K2‑Instruct",
        description="Moonshot AI Kimi‑K2‑Instruct model for code generation and general tasks",
        default_provider="groq"
    ),
    ModelInfo(
        name="DeepSeek V3",
        id="deepseek‑ai/DeepSeek‑V3‑0324",
        description="DeepSeek V3 model for code generation"
    ),
    ModelInfo(
        name="DeepSeek R1",
        id="deepseek‑ai/DeepSeek‑R1‑0528",
        description="DeepSeek R1 model for code generation"
    ),
    ModelInfo(
        name="ERNIE‑4.5‑VL",
        id="baidu/ERNIE‑4.5‑VL‑424B‑A47B‑Base‑PT",
        description="ERNIE‑4.5‑VL model for multimodal code generation with image support"
    ),
    ModelInfo(
        name="MiniMax M1",
        id="MiniMaxAI/MiniMax‑M1‑80k",
        description="MiniMax M1 model for code generation and general tasks"
    ),
    ModelInfo(
        name="Qwen3‑235B‑A22B",
        id="Qwen/Qwen3‑235B‑A22B",
        description="Qwen3‑235B‑A22B model for code generation and general tasks"
    ),
    ModelInfo(
        name="Qwen3‑235B‑A22B‑Instruct‑2507",
        id="Qwen/Qwen3‑235B‑A22B‑Instruct‑2507",
        description="Qwen3‑235B‑A22B‑Instruct‑2507 model for code generation and general tasks"
    ),
    ModelInfo(
        name="Qwen3‑Coder‑480B‑A35B",
        id="Qwen/Qwen3‑Coder‑480B‑A35B‑Instruct",
        description="Qwen3‑Coder‑480B‑A35B‑Instruct model for advanced code generation"
    ),
    ModelInfo(
        name="Qwen3‑32B",
        id="Qwen/Qwen3‑32B",
        description="Qwen3‑32B model for large‑scale code and general tasks"
    ),
    ModelInfo(
        name="SmolLM3‑3B",
        id="HuggingFaceTB/SmolLM3‑3B",
        description="SmolLM3‑3B model for code generation and general tasks"
    ),
    ModelInfo(
        name="GLM‑4.1V‑9B‑Thinking",
        id="THUDM/GLM‑4.1V‑9B‑Thinking",
        description="GLM‑4.1V‑9B‑Thinking model for multimodal code generation with image support"
    ),
    ModelInfo(
        name="OpenAI GPT‑4",
        id="openai/gpt‑4",
        description="OpenAI GPT‑4 via HF Inference Providers",
        default_provider="openai"
    ),
    ModelInfo(
        name="Gemini Pro",
        id="gemini/pro",
        description="Google Gemini Pro via HF Inference Providers",
        default_provider="gemini"
    ),
    ModelInfo(
        name="Fireworks AI",
        id="fireworks‑ai/fireworks‑v1",
        description="Fireworks AI via HF Inference Providers",
        default_provider="fireworks"
    ),
]

def find_model(identifier: str) -> Optional[ModelInfo]:
    """
    Lookup a model by its human name (case‑insensitive) or identifier.

    Returns:
        The matching ModelInfo or None if not found.
    """
    identifier_lower = identifier.lower()
    for m in AVAILABLE_MODELS:
        if m.id == identifier or m.name.lower() == identifier_lower:
            return m
    return None