DocsEditors
Continue
Use the Continue extension for VS Code and JetBrains with your Osarai GPU for chat, edits and agent mode, from one config.yaml.
Continue adds chat, inline edits and an agent to VS Code and JetBrains IDEs. Its extensions and the Continue CLI share one configuration file, so a model you add here works in all of them.
Configure
Open Continue's panel, press the gear icon and choose Open configuration file, or edit the file directly:
~/.continue/config.yaml
name: Osarai
version: 1.0.0
schema: v1
models:
- name: Qwen3.8 27B (Osarai)
provider: openai
model: qwen3.8-27b
apiBase: https://osarai.xyz/v1
apiKey: ${{ secrets.OSARAI_API_KEY }}
roles:
- chat
- edit
- apply
capabilities:
- tool_use
defaultCompletionOptions:
contextLength: 131072
maxTokens: 16384Keep the key in Continue's secrets file next to it:
~/.continue/.env
OSARAI_API_KEY=qck_your_key_hereOn Windows the folder is %USERPROFILE%\.continue.
rolesdecides where the model appears:chatfor the chat panel and agent mode,editandapplyfor inline changes.capabilities: [tool_use]enables agent mode, which reads and edits files with tools.contextLengthis your instance's context window.
Embeddings for @codebase
To search your code by meaning, run qwen3-embedding-8b on a second instance and add it with the embed role:
~/.continue/config.yaml
models:
# … the chat model above …
- name: Qwen3 Embedding 8B (Osarai)
provider: openai
model: qwen3-embedding-8b
apiBase: https://osarai.xyz/v1
apiKey: ${{ secrets.OSARAI_API_KEY }}
roles:
- embedRequests go to the instance whose model matches model. See Embeddings.
Troubleshooting
| Symptom | Fix |
|---|---|
| The model is missing from the model picker | Check that config.yaml parses (Continue shows an error banner) and that roles includes chat |
| Agent mode is greyed out | Add capabilities: [tool_use] to the model |
401 | OSARAI_API_KEY is missing from ~/.continue/.env, or the name differs from the one after secrets. |
503 … not ready yet | The instance is still starting; wait for Ready on the instance page |