Evose

API · Model

Direct call to LLM / Embedding / Reranking · List available models

Model API

Call models directly (bypassing Agent / Workflow orchestration). Suitable for simple scenarios or custom flows.

List Available Models

GET /v1/models
{
  "data": [
    {
      "id": "claude-sonnet-4-6",
      "type": "llm",
      "vendor": "anthropic",
      "context_length": 200000,
      "supports_function_call": true
    },
    {
      "id": "text-embedding-3-large",
      "type": "embedding",
      "vendor": "openai",
      "dimensions": 3072
    }
  ]
}

LLM Invocation

POST /v1/models/llm/chat
{
  "model": "claude-sonnet-4-6",
  "messages": [
    { "role": "system", "content": "You are an assistant." },
    { "role": "user", "content": "1+1=?" }
  ],
  "temperature": 0.2,
  "max_tokens": 100,
  "stream": false
}

OpenAI-compatible format — most clients can be reused directly.

Embedding Invocation

POST /v1/models/embeddings
{
  "model": "text-embedding-3-large",
  "input": ["Hello world", "Bonjour le monde"]
}
{
  "data": [
    { "embedding": [0.012, -0.034, ...], "index": 0 },
    { "embedding": [0.018, -0.022, ...], "index": 1 }
  ],
  "usage": { "tokens": 8, "credits": 0.0001 }
}

Reranking Invocation

POST /v1/models/rerank
{
  "model": "rerank-v3",
  "query": "return policy",
  "documents": [
    "We offer 7-day no-questions-asked returns...",
    "Our hours are Monday to Friday 9am to 6pm...",
    "..."
  ],
  "top_n": 3
}

Next Steps

On this page