Aller au contenu principal
Velqa
← Back to home

Chat / LLM with Velqa — OpenAI-compatible API payable in MAD

Velqa's chat API follows the OpenAI POST /v1/chat/completions format exactly. To use it from existing code, just change the `base_url` to https://api.velqa.dev/v1 and use your sk-... key. You pay in dirhams, no international card required.

Minimal request (curl)

curl https://api.velqa.dev/v1/chat/completions \
  -H "Authorization: Bearer sk-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k2.6",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user", "content": "Explain RAG in one sentence."}
    ],
    "max_tokens": 200
  }'

Minimal request (Python, openai SDK)

from openai import OpenAI

client = OpenAI(base_url="https://api.velqa.dev/v1", api_key="sk-...")

resp = client.chat.completions.create(
    model="kimi-k2.6",
    messages=[
        {"role": "system", "content": "You are a concise assistant."},
        {"role": "user", "content": "Explain RAG in one sentence."},
    ],
    max_tokens=200,
)

print(resp.choices[0].message.content)

Request shape

  • model — the model ID (see models) or velqa-auto to let Velqa choose (see auto-router).
  • messages — the conversation, a list of {"role": ..., "content": ...} objects. Roles are system, user, assistant (and tool for tool calling).
  • max_tokens — caps the number of generated output tokens. Useful to control cost and avoid exceeding the context window of a small-context model.
  • temperature, top_p, stop… — the usual OpenAI parameters are supported.

Response shape

{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "kimi-k2.6",
  "choices": [
    {
      "index": 0,
      "message": {"role": "assistant", "content": "..."},
      "finish_reason": "stop"
    }
  ],
  "usage": {"prompt_tokens": 42, "completion_tokens": 18, "total_tokens": 60}
}

The generated text is in choices[0].message.content. The usage field reports the billed tokens (input + output).

Choosing a model

  • Fast and cheap: glm-4.7-flash, deepseek-v4-flash.
  • Coding and agentic: kimi-k2.6, minimax-m3, glm-5.2, deepseek-v4-pro, qwen3.7-max.
  • Don't want to choose? Use velqa-auto.

Advanced coding models require a Dev plan or higher. See models and plans.

Going further