Model API

The DGrid Model API consolidates DGrid's model-facing endpoints into a single reference section. It covers Gemini-compatible, OpenAI-compatible, Claude-compatible, and realtime interfaces so teams can choose the protocol that matches their application stack while keeping DGrid as the unified backend.

Looking for tool-specific onboarding instead of endpoint reference? See Integration Tutorials.

AudioSpeech generation, transcription, translation, and Gemini-native audio understanding.ChatClaude Messages, Gemini multimodal chat, OpenAI Chat Completions, and Responses API.CompletionsLegacy OpenAI-compatible text completion requests for prompt-based generation.EmbeddingsOpenAI-compatible and Gemini-native embedding endpoints for retrieval and semantic search.ImagesGenerate and edit images through Gemini, OpenAI image APIs, or Qwen-compatible endpoints.ModerationsRun content safety checks against text inputs before or after generation.RealtimeOpenAI-compatible websocket sessions for low-latency text and audio conversations.

Quickstart

Use the OpenAI-compatible chat endpoint for the fastest first integration:

curl https://api.dgrid.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $DGRID_API_KEY" \
  -d '{
    "model": "openai/gpt-4o",
    "messages": [
      {
        "role": "user",
        "content": "Hello from DGrid."
      }
    ]
  }'

Authentication Formats

OpenAI-Compatible

Authorization: Bearer <DGRID_API_KEY>
Content-Type: application/json

Use this format for Chat Completions, Responses, Completions, Embeddings, Images, Moderations, audio speech endpoints, and realtime session creation.

Claude-Compatible

Authorization: Bearer <DGRID_API_KEY>
anthropic-version: 2023-06-01
Content-Type: application/json

Use this format for the POST /v1/messages interface.

Gemini-Compatible

Authorization: Bearer <DGRID_API_KEY>
Content-Type: application/json

Use this format for generateContent and embedContent requests.

Endpoint Families

Family	Primary Base Path	Best For
Audio	/v1/audio/* and /v1/models/*:generateContent	Speech synthesis, transcription, translation, and multimodal audio understanding
Chat	/v1/messages, /v1/chat/completions, /v1/responses, /v1/models/*:generateContent	Conversational apps, multimodal assistants, tool calling
Completions	/v1/completions	Prompt-only text generation and legacy integrations
Embeddings	/v1/embeddings, /v1/models/*:embedContent	Search, clustering, retrieval, ranking
Images	/v1/images/*, /v1/chat/completions, /v1/models/*:generateContent	Image generation and editing across multiple model formats
Moderations	/v1/moderations	Content safety filtering and policy checks
Realtime	/v1/realtime, /v1/realtime/sessions	Low-latency streaming conversations over websocket

Best Practices

1. Store the DGrid API key on the server side and never expose it in public clients or repositories.
2. Prefer the OpenAI-compatible format when you want the broadest SDK compatibility with minimal migration effort.
3. Use provider-native formats only when you need provider-specific payload features such as Gemini multimodal parts or Claude tool payloads.
4. Keep request and response examples in sync with your target model family, because parameter names differ between OpenAI, Claude, and Gemini styles.