Skip to main content

Sandbox code execution

Feature Request

ChatGPT can run my code and verify if it is correct. Would love tool use/being able to have t3 run code. Especially for models like claude.

Makes it super easy to demo and test if the code is correct right in the app. Especially for frontend

Also great to build demo apps/visualizations

Status: Gathering Interest6 comments

Log in to comment and vote

Comments6

  • Apricot Honey

    •

    Merged request

    Code Interpreeter using WebContainers

    Add code interpreter support, make it local using the WebContainers API. Could make it work for both Python and Javascript, python might be more finicky since support is still in beta but Javscript should be easy.

  • Plum Grove

    Absolute must have feature. You cannot rely on an LLM to do math unless it can verify the calculation in code.

  • Aqua Toast

    This feature would mean me sticking with this subscription after this first trial month. Claude supports this in the API it seems:

    https://docs.anthropic.com/en/docs/agents-and-tools/tool-use/code-execution-tool

  • Amethyst Lemon

    Openai code interpreter cost 3cents/session via api. A session is one user, up to an hour. This would match well to one t3 credit.

  • Fuchsia Biscuit

    Coolest way to do this:

    https://webcontainers.io/ai
    Probably a pain though

    • Aqua Reindeer

      it isn’t, webcontainers are relatively easy imo