Skip to main content
POST
/
v1
/
chat
/
completions
Generative LLM
curl --request POST \
  --url https://api.classer.ai/v1/chat/completions \
  --header 'Authorization: Bearer <token>'
{
  "object": "list",
  "data": [
    {
      "id": "deepseek-ai/DeepSeek-V4-Flash",
      "hugging_face_id": "deepseek-ai/DeepSeek-V4-Flash",
      "name": "DeepSeek: V4 Flash",
      "input_modalities": ["text"],
      "output_modalities": ["text"],
      "quantization": "bf16",
      "context_length": 1000000,
      "max_output_length": 128000,
      "pricing": {
        "prompt": "0.0000001",
        "completion": "0.00000028",
        "image": "0",
        "request": "0",
        "input_cache_read": "0"
      },
      "supported_features": ["tools", "json_mode"]
    }
  ]
}

Documentation Index

Fetch the complete documentation index at: https://docs.classer.ai/llms.txt

Use this file to discover all available pages before exploring further.

Classer exposes an OpenAI-compatible generation surface so existing tools and SDKs work unchanged. Point any client at https://api.classer.ai/v1 with your Classer API key and call chat/completions, responses, or models exactly as you would against OpenAI.
Drop-in replacement for OpenAI clients. Set base_url to https://api.classer.ai/v1 and api_key to your Classer key — no other code changes needed.

​
Get an API key

Create one at classer.ai/api-keys. You can configure per-key logging (off by default for inputs and outputs) and an optional retention window. The key value is shown once at creation — copy it immediately.

​
Available models

ModelContextInput ($/1M)Cache read ($/1M)Output ($/1M)Notes
deepseek-ai/DeepSeek-V4-Flash1M$0.10–$0.28Open-weights, cost-sensitive workloads
Qwen/Qwen3.6-35B-A3B256K$0.15$0.05$1.00Vision, tool use, structured outputs
moonshotai/Kimi-K2.6256K$0.70–$4.00Reasoning, vision, tool use
Pricing applies to both /v1/chat/completions and /v1/responses. Use GET /v1/models for the live machine-readable catalog including per-model supported_features, supported_sampling_parameters, quantization, and modalities.

​
Endpoints

MethodEndpointDescription
POST/v1/chat/completionsOpenAI Chat Completions API
POST/v1/responsesOpenAI Responses API
GET/v1/modelsList available generative models

​
Chat completions

POST /v1/chat/completions accepts the OpenAI Chat Completions request shape verbatim and returns the upstream response unchanged. Both streaming (stream: true) and non-streaming requests are supported. Usage tokens are populated in the response for both modes — the proxy automatically negotiates stream_options.include_usage where needed so billing is exact.

​
Responses

POST /v1/responses is the OpenAI Responses API. Same models, same pricing as chat completions; the request and response shapes match OpenAI’s spec.

​
List models

GET /v1/models returns the live deployment list with full OpenRouter-style metadata: pricing, context length, max output length, quantization, input/output modalities, supported sampling parameters, and supported features per model. 
curl https://api.classer.ai/v1/models \
  -H "Authorization: Bearer $CLASSER_API_KEY"

{
  "object": "list",
  "data": [
    {
      "id": "deepseek-ai/DeepSeek-V4-Flash",
      "hugging_face_id": "deepseek-ai/DeepSeek-V4-Flash",
      "name": "DeepSeek: V4 Flash",
      "input_modalities": ["text"],
      "output_modalities": ["text"],
      "quantization": "bf16",
      "context_length": 1000000,
      "max_output_length": 128000,
      "pricing": {
        "prompt": "0.0000001",
        "completion": "0.00000028",
        "image": "0",
        "request": "0",
        "input_cache_read": "0"
      },
      "supported_features": ["tools", "json_mode"]
    }
  ]
}

​
Errors

StatusCodeCause
401INVALID_API_KEYMissing or invalid Authorization: Bearer … header
400invalid_request_errormodel not in the allow-list (response includes the list of valid models)
400invalid_request_errormodel field missing entirely
402INSUFFICIENT_BALANCEAccount balance is below the grace buffer; top up to resume
503–Generate gateway not configured / unreachable