Reply Better AIAdd to Chrome
Local AI · Ollama · LM Studio

Use Ollama to write and reply on any website

Reply Better AI connects to the Ollama or LM Studio server on your own machine. Select a message, get a reply. Type a draft, get a cleaner version. Nothing goes to the cloud.

Add to Chrome — it's freeFirefox

Why run it locally

Private. Your drafts and the messages you reply to stay on your computer. There is no account and no cloud provider in between.

Free and unlimited. You’re running the model, so there is no quota and no bill.

Your model, your choice. Use whatever model you already like in Ollama or LM Studio, and switch any time.

Set it up with Ollama

  1. Install Ollama from ollama.com/download and pull a model:
ollama pull llama3.2
ollama serve
  1. Allow the extension to reach the server. Ollama only accepts requests from localhost by default, so browser extensions get blocked:
OLLAMA_ORIGINS="chrome-extension://*,moz-extension://*" ollama serve

On the macOS app, run launchctl setenv OLLAMA_ORIGINS "chrome-extension://*,moz-extension://*" and restart Ollama. On Windows, add OLLAMA_ORIGINS with the same value in Environment Variables and restart Ollama. Chrome often works without this step. Firefox always needs it.

  1. In the extension, open Settings → Engine → Local (Ollama / LM Studio), click the Ollama preset and pick a model. The status line should show “Connected”.

The first request after starting the server can take a few seconds while the model loads. After that it’s fast.

Set it up with LM Studio

  1. Open the Developer tab, download a model and load it.
  2. Start the server. It runs on http://localhost:1234 by default.
  3. Turn on Enable CORS in the server settings. It’s off by default, and the server rejects every browser request until you turn it on.
  4. In the extension, pick Local, click the LM Studio preset and choose your model.

If it doesn’t connect

What you see What to do
Can’t reach the server Check the server is running and the URL ends in /v1. Turn on CORS in LM Studio, or set OLLAMA_ORIGINS for Ollama.
Reachable, but no models Pull a model in Ollama, or load one in LM Studio.
First request is slow Ollama is loading the model. The next ones are fast.

Questions

Do I need an API key?

No. The extension talks to your local server without a key. If you turned on an API key in LM Studio, turn it off for this.

Which models work?

Any chat model your Ollama or LM Studio server can run. The extension lists the models you have installed and you pick one. Smaller models answer faster, bigger ones write better.

Does it work in Firefox?

Yes, but Firefox needs the OLLAMA_ORIGINS step below. Chrome usually works without it.

Does it work offline?

Yes. Once the model is downloaded, your server and the extension don't need the internet.

Can I use a server on another port or machine?

Pick Custom in the Local server settings and enter its base URL. It has to be OpenAI-compatible and end in /v1, which covers llama.cpp, vLLM and LiteLLM too.