Chat Completions

POST https://asklethe.ai/v1/chat/completions accepts the same request shape as OpenAI’s Chat Completions API. Your request body is forwarded as-is (after LETHE resolves model to the upstream model), so any parameter the OpenAI API accepts, temperature, max_tokens, tools, tool_choice, stream, etc., passes through.

curl https://asklethe.ai/v1/chat/completions \
  -H "Authorization: Bearer $LETHE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "lethe",
    "messages": [{ "role": "user", "content": "Say hi in five words." }]
  }'

const res = await client.chat.completions.create({
  model: "lethe",
  messages: [{ role: "user", content: prompt }],
});

resp = client.chat.completions.create(
    model="lethe",
    messages=[{"role": "user", "content": prompt}],
)

Streaming

Set "stream": true to receive server-sent events, same as the OpenAI API. Model names in streamed chunks are rewritten to the public lethe / lethe-vision id, you will never see an upstream provider name in the stream.

Vision input

To send image or video content parts, use model: "lethe-vision". Sending vision content to lethe (text-only) returns an invalid_model error. See Models.

Statelessness

LETHE does not store or merge conversation history between calls. Send the full messages array on every request.

Response headers

Every response includes:

Header	Meaning
`X-Lethe-Remaining`	Requests left in your current quota window.
`X-Lethe-Limit`	Your total quota for the window (3 guest, 50 wallet-linked).

See Rate Limits & Quotas for details, and Errors for failure responses.

⌘I

Introduction

User Guide

API Reference

Integrations

Roadmap

Support

Legal

Chat Completions

Streaming

Vision input

Statelessness

Response headers

​Streaming

​Vision input

​Statelessness

​Response headers

Streaming

Vision input

Statelessness

Response headers