Home / Integrations / Continue

Use Continue with TurboLLM

Continue is a free, open-source VS Code and JetBrains extension built to support local, self-hosted OpenAI-compatible servers like TurboLLM. Point it at your local endpoint and you get chat, edit, and autocomplete backed entirely by models running on your own machine.

Before you start

TurboLLM exposes an OpenAI-compatible API, so Continue talks to it the same way it would talk to any hosted provider — just with a local apiBase. You need two things: your Server URL and (optionally) an API key.

  1. Load a model — or don't

    Load a model in TurboLLM, or skip this step. Most clients trigger an on-demand load on the first request, so Continue will bring a model up when it sends its first message.

  2. Grab your Server URL and API key

    Copy the Server URL (for example http://localhost:6996). If you've turned on Require API key under Settings → Network, create a key from Developer → API Keys → Create. If Require API key is off, you can leave the key out entirely.

  3. Install Continue

    Install Continue from the VS Code Marketplace. Continue also ships a JetBrains plugin if you work in IntelliJ, PyCharm, or another JetBrains IDE.

  4. Add a model block to Continue's config

    Open Continue's config with the gear icon next to Local Config in its sidebar, or edit ~/.continue/config.yaml directly (%USERPROFILE%\.continue\config.yaml on Windows). Add the model block shown below.

  5. Save and select the model

    Save the file — Continue reloads automatically — then pick TurboLLM (local) from its model dropdown.

Config

Paste this into your config.yaml, adjusting the URL and key to match your setup:

models:
  - name: TurboLLM (local)
    provider: openai
    model: local              # TurboLLM fuzzy-matches any name and loads it on demand
    apiBase: http://localhost:6996/v1
    apiKey: <your TurboLLM API key, or omit if Require API key is off>
    roles:
      - chat
      - edit
      - autocomplete
Why model: local works

TurboLLM fuzzy-matches whatever model name a client sends and loads it on demand, so local is a safe placeholder. Continue's three roles — chat, edit, and autocomplete — all route to the same TurboLLM endpoint.

API key required?

The apiKey line is only needed if you've enabled Require API key under Settings → Network. With that off, omit the line entirely.

Recommended models

For editing and autocomplete inside your IDE, a coding-focused model works best. See the coder picks in the model catalog for options to load in TurboLLM.

Previous← Hermes Agent