Which AI models can you use in Mel?

Mel – the agentic terminal (openmel.dev) works with Claude, GPT, Gemini, DeepSeek and Qwen, so you can use US and Chinese models in one tool. You can pick the model for a task yourself, or leave it on Auto and let Mel choose for each turn. You pay either with your own API key, which is free in Mel, or with hosted credits on a paid plan. This page explains the choices and how to use them well.

Which models does Mel support?

Mel works with these model families:

This is not a ranking. Models change often, and the best model for your project is the one that does best on your own tasks.

How do you choose a model in Mel?

You have two ways.

1. Pick it yourself. Choose a model for the conversation when you know what you want, for example the model your team already trusts for reviews.
2. Leave it on Auto. Mel chooses a model for each turn, so you do not have to think about it for routine work.

Tip: use a different model to review a change than the one that wrote it. A second model is more likely to question what the first one took for granted.

Do you need your own API key?

No, but you can use one. There are two ways to pay for model use in Mel:

• Your own API key. Mel is free with your own key on Mac, Windows and Linux. Your provider bills the usage on its own price and its own bill. Read the provider's terms, including what a free tier may do with what you send.
• Hosted credits. Pro ($20 a month) and Max ($200 a month) include hosted credits, so you do not need a provider key.

Why use more than one model?

No single model is best at every step. Planning, writing code, fixing a failing test and reviewing a diff can each suit a different model. With several models in one tool, you change the model, not the tool: the task, the history and your approvals stay in one place.

A simple test:

1. Give Mel one real task with a clear finish line, such as "make this test pass".
2. Run it on one model, then on a second.
3. Compare the result, the time it took and the cost, and keep the better one for that kind of step.

Does the model change what Mel can do?

Mel's agents work the same way whatever model you pick: they plan the work, edit files, run the build and the tests, and fix what fails. Configurable approvals keep you in control of commands and edits, whichever model is doing the work.

What about teams?

For a team, the Business plan is $50 a seat a month, two seats minimum, with shared credits and one invoice, so you do not track a separate subscription for each person. See pricing.

Frequently asked questions

Can I use Claude and GPT in the same tool? Yes. Mel works with Claude, GPT, Gemini, DeepSeek and Qwen.

What does Auto do? Auto lets Mel choose a model for each turn, so you do not pick one every time.

Is Mel free? Yes, with your own API key. Your provider's usage charges apply.

Try Mel

Download Mel for free and try two models on the same task. To see model choice on your team's own repo, write to hello@openmel.dev and we will set up a short demo.

Mel – the agentic terminal · openmel.dev. The agentic team for coding.