Models

Retrieve information about available models.

List Models

GET /v1/models

Example Request

curl https://api.solidrust.ai/v1/models \
  -H "Authorization: Bearer YOUR_API_KEY"

Response

{
  "object": "list",
  "data": [
    {
      "id": "vllm-primary",
      "object": "model",
      "created": 1704067200,
      "owned_by": "solidrust",
      "permission": [],
      "root": "vllm-primary",
      "parent": null
    },
    {
      "id": "bge-m3",
      "object": "model",
      "created": 1704067200,
      "owned_by": "solidrust",
      "permission": [],
      "root": "bge-m3",
      "parent": null
    }
  ]
}

Available Models

Chat Models

Model ID	Description	Context Length	Best For
vllm-primary	Recommended alias (currently Qwen3-4B)	8192	All chat tasks - use this
qwen3-4b	Direct model reference	8192	When you need a specific model

Use vllm-primary

Always use vllm-primary in your code. This alias automatically routes to our best available model and ensures your integration survives model upgrades without code changes.

Embedding Models

Model ID	Dimensions	Max Input	Best For
bge-m3	1024	8192 tokens	Semantic search, RAG

Model Selection

• Chat completions: Use vllm-primary
• Embeddings: Use bge-m3

Failover Behavior

When local GPU infrastructure is unavailable, requests automatically route to cloud providers:

Primary	Failover Chain
vllm-primary	OpenAI GPT-4o-mini → Claude Haiku

This ensures high availability while maintaining API compatibility. You can detect failover by checking the model field in responses - it will indicate which model actually served the request.