Developers

Lexible API

Every Lexible machine exposes an OpenAI-compatible API. Your data never leaves the machine: requests go straight to it, over its own TLS certificate, through the blind relay when you are off-site. Point any OpenAI client library at the machine's base URL and use a Lexible API key.

Base URL and keys

Sign in to see your machines' addresses. Addresses look like https://acme-law.on.lexible.ai.

Keys look like lxk_xxxxxxxx_…. Organization admins create them on the organization page (valid on every machine in the organization) or on one machine's Developers page (valid there only). Each key has scopes and a per-minute rate limit. Keys are stateless: they never read or write anyone's conversations.

Scope	Allows
chat	POST /chat/completions (streaming with stream: true)
models	GET /models
embeddings	POST /embeddings
images	POST /images/generations (501 until the image engine ships)
status	GET /status — load, model readiness, capacity

Examples

curl https://your-machine.on.lexible.ai/api/v1/chat/completions \
  -H "Authorization: Bearer $LEXIBLE_API_KEY" -H "Content-Type: application/json" \
  -d '{"model": "default", "messages": [{"role": "user", "content": "Summarise this contract clause: …"}]}'

from openai import OpenAI
client = OpenAI(base_url="https://your-machine.on.lexible.ai/api/v1", api_key=os.environ["LEXIBLE_API_KEY"])
reply = client.chat.completions.create(model="default", messages=[{"role": "user", "content": "Hello"}])

import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://your-machine.on.lexible.ai/api/v1", apiKey: process.env.LEXIBLE_API_KEY });
const out = await client.embeddings.create({ model: "default", input: ["invoice", "receipt"] });

Behaviour

• Errors use the OpenAI shape: {"error": {"message", "type"}}. 401 bad key, 403 missing scope, 429 over the per-minute limit (with Retry-After).
• model may be default or any id from GET /models.
• Capacity: GET /status reports ok, busy or critical; back off when it is not ok.
• The interactive reference (OpenAPI) is served by each machine at https://your-machine.on.lexible.ai/docs.
• Multiple machines: today each key works on every machine in the organization; pick one base URL per integration. Automatic routing across machines is on the roadmap.