Support for Foundry Local

[ShahriarKh]

Is Foundry Local supported? If yes, where is it? If no, are there plans to add it?
https://www.foundrylocal.ai/

[ShahriarKh]

Okay I found it. It's not listed as a provider, but we can use it since it's openai compatible:

name: Local Config
version: 1.0.0
schema: v1
models:
  - name: foundry
    provider: openai
    model: AUTODETECT
    apiBase: http://localhost:64337/v1
    useLegacyCompletionsEndpoint: true

[xXMrNidaXx]

Foundry Local uses an OpenAI-compatible API, so you can use it with Continue today via the generic OpenAI provider.

Configuration:

models:
  - name: foundry-local
    provider: openai
    model: phi-4  # or whatever model you have loaded
    apiBase: http://localhost:5272/v1
    apiKey: not-needed  # Foundry Local does not require auth
    roles:
      - chat
      - edit

Steps:

1. Start Foundry Local (it runs on port 5272 by default)
2. Load your model in Foundry
3. Add the config above to your Continue config
4. Select the model in Continue

Verify it works:

curl http://localhost:5272/v1/models

Tips:

• Foundry Local supports multiple models simultaneously
• Check which model is loaded before configuring
• For tool calling, ensure your Foundry model supports it

If you want native provider support:
The generic OpenAI provider works well, but a dedicated Foundry provider could add:

• Model auto-discovery from Foundry catalog
• Download progress in Continue UI
• Foundry-specific features

Might be worth a feature request if there is enough interest!

We test various local inference backends at Revolution AI — Foundry Local + Continue works smoothly via OpenAI compat.

[nwillems94]

Has anyone been able to get this to work with autocomplete? I have the following config:

  - name: Qwen2.5-coder
    provider: openai
    model: qwen2.5-coder-1.5b-instruct-generic-cpu:4
    apiBase: http://localhost:5272/v1
    useLegacyCompletionsEndpoint: false
    roles:
      - autocomplete
      - edit
      - chat

No error card is displayed, but in the developer tools console I see

The operation was aborted. {"context":"llm_fetch","url":"http://localhost:5272/v1/chat/completions","method":"POST","model":"qwen2.5-coder-1.5b-instruct-generic-cpu:4","provider":"openai"}

Chat functionality works fine and I have tried setting the autocomplete timeout to the maximum of 5000ms.
Setting useLegacyCompletionsEndpoint: true results in a HTTP 404 error so I don't think that is the issue.

[ShahriarKh]

Does this help https://docs.continue.dev/customize/deep-dives/autocomplete ?

[nwillems94]

No, I tried the "Autocomplete Not Working – How to Fix It" trouble shooting steps. I also tried adding this to the YAML config

    autocompleteOptions:
      disable: false
      modelTimeout: 5000

[ShahriarKh]

Maybe try with a smaller model, like qwen2.5-coder-0.5b, or nate/instinct (ollama, not foundry) and see if the issue persists