POST /v1/messages

Le principal point final compatible avec Anthropic. Il prend en charge le streaming, la mise en cache rapide, la vision, l'utilisation d'outils et la réflexion étendue lorsqu'il est pris en charge par le modèle sélectionné.

Authentification

x-api-key: mg_live_...\nanthropic-version: 2023-06-01\ncontent-type: application/json

Demande minimale

curl https://api.model-gate.com/v1/messages \
  -H "x-api-key: mg_live_..." \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4.8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Demande avancée

{
  "model": "claude-opus-4.8",
  "max_tokens": 4096,
  "system": "You are a coding assistant.",
  "thinking": {"type": "enabled", "budget_tokens": 8000},
  "messages": [{"role": "user", "content": "Find the bug in this code."}],
  "tools": [{
    "name": "search_files",
    "description": "Search project files",
    "input_schema": {"type": "object", "properties": {"q": {"type": "string"}}}
  }]
}

Vision

{
  "model": "claude-opus-4.8",
  "max_tokens": 1024,
  "messages": [{
    "role": "user",
    "content": [
      {"type": "text", "text": "What is shown in this image?"},
      {"type": "image", "source": {"type": "base64", "media_type": "image/png", "data": "..."}}
    ]
  }]
}

Mise en cache rapide

{
  "messages": [{
    "role": "user",
    "content": [
      {"type": "text", "text": "Large reusable context...", "cache_control": {"type": "ephemeral"}},
      {"type": "text", "text": "New question"}
    ]
  }]
}

Streaming

Ensemble "stream": true. La réponse est renvoyée sous forme d'événements envoyés par le serveur.

Réponse

{
  "id": "msg_...",
  "type": "message",
  "role": "assistant",
  "model": "claude-opus-4.8",
  "content": [{"type": "text", "text": "Hello!"}],
  "stop_reason": "end_turn",
  "usage": {"input_tokens": 12, "output_tokens": 8}
}

La structure des requêtes et des réponses suit l'API Anthropic Messages. Les SDK Anthropic existants nécessitent normalement uniquement le base_url changement.