Quickstart

Do I call CompresLM or my LLM first?

Call CompresLM only. You do not send a separate request to OpenAI/Anthropic first. CompresLM is a drop-in proxy: your app talks to https://compreslm.com/v1, we compress what we can, then we forward the request to your provider with your key. The model reply comes back through us in the same OpenAI-compatible shape.

We never store your prompts, completions, or provider key. We only record token counts (before/after) for your dashboard and bill — see Privacy Policy.

Create an account and a token

Sign up at /app, verify your email, then click Generate token. Copy it once — it is shown only at creation time. This token goes in the Authorization header on every request; it identifies your account, not your LLM provider.

Point your client at CompresLM

Change two things in your existing OpenAI-compatible client: the base URL (point it at CompresLM) and add one extra header carrying your own provider key. Everything else — model names you're used to, message format, streaming, tool calls — stays the same.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://compreslm.com/v1",
    api_key="YOUR_COMPRESLM_TOKEN",
    default_headers={"X-LLM-API-Key": "YOUR_PROVIDER_KEY"},
)

resp = client.chat.completions.create(
    model="gpt-4o-mini",  # any model name your provider accepts
    messages=[{"role": "user", "content": "Summarize nothing — just say hi."}],
)
print(resp.choices[0].message.content)

cURL

curl https://compreslm.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_COMPRESLM_TOKEN" \
  -H "X-LLM-API-Key: YOUR_PROVIDER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role":"user","content":"Hello"}]
  }'

JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://compreslm.com/v1",
  apiKey: "YOUR_COMPRESLM_TOKEN",
  defaultHeaders: { "X-LLM-API-Key": "YOUR_PROVIDER_KEY" },
});

const r = await client.chat.completions.create({
  model: "gpt-4o-mini", // any model name your provider accepts
  messages: [{ role: "user", content: "Hello" }],
});

What you get back

A standard chat-completion response from your provider — same fields you already parse (choices, usage, etc.). CompresLM shortens the prompt by default. With compress_output: true it also reduces completion tokens — still a normal model answer in the usual response shape.

Choose your model

Send any model name your provider accepts — the exact string you already pass today (gpt-4o-mini, gpt-4.1, o3, claude-sonnet-4-6, claude-opus-4-1, …). There is no supported-model list to pick from: we forward the name to your provider unchanged, on your key. Compression is on by default and does not change which model answers you.

What to send where

This is the one idea worth five minutes. A request carries two different things, and telling them apart is what makes compression safe:

Put both in one big user message and it still works — we fall back to reading the tail of the message as the question. But that is a guess, and a wrong guess is how compression drops something you needed. Separating them removes the guess.

The same request, both ways

# Guessing — document and question in one message
messages=[{"role": "user", "content": contract + "\n\nWhat is the notice period?"}]

# Told — the document is context, the question is the message
messages=[{"role": "user", "content": "What is the notice period?"}],
extra_body={"context": contract}

Python SDKs pass it through extra_body; in JavaScript and cURL it is just another top-level field:

curl https://compreslm.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_COMPRESLM_TOKEN" \
  -H "X-LLM-API-Key: YOUR_PROVIDER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "context": "...the contract, the report, the retrieved chunks...",
    "messages": [{"role":"user","content":"What is the termination notice period?"}]
  }'

Shortening the answer is a separate, opt-in flag — compress_output, on the profiles page.