modelis-openai

Use Modelis in Aider, Cline, Continue, or any OpenAI SDK — with one base_url and zero migration.

Modelis auto-routes every request to the best model (GPT / Claude / Gemini) and bills a flat, predictable price per call — not per token. It's distributed through RapidAPI, which uses an X-RapidAPI-Key header instead of the Authorization: Bearer header that coding tools expect.

This is the ~120-line bridge: a zero-dependency local proxy that speaks the OpenAI wire protocol on 127.0.0.1, rewrites the auth, and forwards to Modelis. Point any OpenAI-compatible tool at it and you're done.

your tool ──OpenAI(Bearer)──▶ modelis-openai (localhost) ──X-RapidAPI-Key──▶ Modelis on RapidAPI ──▶ best model

Quickstart

You need a RapidAPI key. Subscribe to the free tier here and copy your key from the Endpoints tab: 👉 https://rapidapi.com/chenxiao5580/api/modelis-auto-chat

Run with Node (18+):

npx modelis-openai
# or:  node modelis-openai.mjs

Run with Docker (no Node needed):

git clone https://github.com/modelishub/modelis-openai
cd modelis-openai
docker build -t modelis-openai .
docker run --rm -p 8787:8787 modelis-openai

Then point your tool at:

| Setting | Value | |-----------|--------------------------------| | Base URL | http://127.0.0.1:8787/v1 | | API key | your RapidAPI key | | Model | modelis-auto |

That's it. Verify it's alive:

curl http://127.0.0.1:8787/health

Tool recipes

Aider

export OPENAI_API_BASE=http://127.0.0.1:8787/v1
export OPENAI_API_KEY=<your-rapidapi-key>
aider --model openai/modelis-auto

Cline / Roo Code (VS Code)

• API Provider: OpenAI Compatible
• Base URL: http://127.0.0.1:8787/v1
• API Key: your RapidAPI key
• Model ID: modelis-auto

Continue (~/.continue/config.yaml)

models:
  - name: Modelis
    provider: openai
    model: modelis-auto
    apiBase: http://127.0.0.1:8787/v1
    apiKey: <your-rapidapi-key>

Any OpenAI SDK

from openai import OpenAI
client = OpenAI(base_url="http://127.0.0.1:8787/v1", api_key="<your-rapidapi-key>")
print(client.chat.completions.create(
    model="modelis-auto",
    messages=[{"role": "user", "content": "Hello"}],
).choices[0].message.content)

import OpenAI from "openai";
const client = new OpenAI({ baseURL: "http://127.0.0.1:8787/v1", apiKey: "<your-rapidapi-key>" });
const r = await client.chat.completions.create({
  model: "modelis-auto",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(r.choices[0].message.content);

Cursor: Cursor sends requests from its own servers, so a localhost endpoint won't reach this proxy. Use Cursor only if you expose the proxy on a public URL you control. The local setup above is for tools that call the API from your machine (Aider, Cline, Continue, scripts).

How it works

Streaming (stream: true) is piped straight through, so token-by-token output in your tool works exactly as with the OpenAI API. The proxy:

1. reads the key from Authorization: Bearer <key> (or MODELIS_RAPIDAPI_KEY),
2. rewrites the request model to modelis-auto (configurable),
3. forwards to the RapidAPI gateway with X-RapidAPI-Key / X-RapidAPI-Host,
4. streams the response back unchanged.

It also answers GET /v1/models and GET /health so tools that probe on startup don't error.

Configuration

All optional — sensible defaults work out of the box.

| Env var | Default | Purpose | |------------------------------|--------------------------------------|---------| | MODELIS_PORT | 8787 | Local listen port | | MODELIS_HOST | 127.0.0.1 | Local bind address (0.0.0.0 in Docker) | | MODELIS_RAPIDAPI_KEY | (unset) | Fallback key if your tool can't send one | | MODELIS_MODEL | modelis-auto | Model sent upstream; set "" to pass the tool's model through | | MODELIS_RAPIDAPI_HOST | modelis-auto-chat.p.rapidapi.com | RapidAPI gateway host (from your Code Snippet) | | MODELIS_UPSTREAM_PATH | /v1/chat/completions | Upstream endpoint path |

Privacy & trust

• Runs entirely on your machine; the only outbound connection is to the RapidAPI gateway over HTTPS.
• It does not log, store, or inspect message contents — it only swaps the auth header and forwards bytes.
• ~120 lines, no third-party dependencies. Read it.

Develop

node --test     # unit + in-process integration tests, zero deps

License

MIT