ConfigurationSecurity & Privacy
LLM Providers
How third-party LLM providers are used
LLM Providers
How Third-Party Models Are Used
Burg AI sends your code to the LLM provider you choose using your API key.
We support:
- OpenAI (GPT-4 series)
- Anthropic (Claude 3 series)
- OpenRouter (multiple models)
- Google Gemini
What Data Is Sent
To the LLM provider:
| Sent | Not Sent |
|---|---|
| PR diff | Full repository |
| Changed file content | Unchanged files |
| File paths | User identity |
| Custom instructions | API keys of other users |
BYOK Implications
With BYOK:
- You control costs — Usage is billed to your account
- You control data policies — Subject to your agreement with the provider
- You can audit usage — Check your provider dashboard
Provider Data Policies
Each provider has their own data retention policy:
| Provider | Training on API Data | Data Retention |
|---|---|---|
| OpenAI | No (API data not used for training) | 30 days for abuse monitoring |
| Anthropic | No | Ephemeral |
| Varies by model/region | Varies | |
| OpenRouter | Pass-through | Depends on underlying model |
Check each provider's terms for current policies.
Burg AI's Role
Burg AI:
- Routes your request to the provider
- Never stores the response after posting
- Never trains on your code
- Cannot access other users' data via your key