Devotel AI

An OpenAI-compatible API for Kimi K3, hosted by Devotel. If your code already talks to OpenAI, point it at a new base URL and give it a Devotel key — nothing else changes.

https://api.devotel.com/v1

from openai import OpenAI
 
client = OpenAI(api_key="dvt-...", base_url="https://api.devotel.com/v1")

What you get

• POST /v1/chat/completions — streaming and non-streaming, with tool calling
• GET /v1/models — the models available to your key
• Keys you create and revoke yourself at chat.devotel.com/keys 
• A usage dashboard with per-key token counts and cost

Two things to know up front

Kimi K3 is a reasoning model. Responses include a reasoning_content field alongside content, and the tokens spent thinking are billed as output tokens. See Chat Completions before you budget or set max_tokens.

Privacy. For API traffic the gateway stores token counts and metadata only — never your prompts or the model’s replies. Conversations you have in the chat app  are stored, because they are your history.

Where to go next

• Quickstart — a working request in curl, Python or JavaScript
• Tool Calling & Agents — a complete agent loop in Python and TypeScript
• Rate Limits & Errors — every error code the API returns