How to Configure BYOLLM
BYOLLM (Bring Your Own LLM) lets you point aprity at your own LLM endpoint instead of aprity's default managed LLM infrastructure.
:::info Intelligence plan only BYOLLM is available on the Intelligence plan only. It is not included on the Documentation or Trial plans. :::
Prerequisites
- An Intelligence plan with BYOLLM enabled for your aprity tenant. Contact support@aprity.ai if it is not visible in your settings.
- An active endpoint and API key from a supported provider.
- A model with a 1M-token context window (aprity targets 1M-context models exclusively).
Supported providers
| Provider | Requirements |
|---|---|
| Anthropic (Claude) | Anthropic API key, 1M-context model |
| OpenAI | OpenAI API key, 1M-context model |
| Azure OpenAI | Azure subscription, deployed 1M-context model, API key and endpoint |
| xAI (Grok) | xAI API key, 1M-context model |
How to Set Up BYOLLM
BYOLLM configuration is handled by the aprity support team to ensure secure key management.
- Contact support@aprity.ai with your BYOLLM request.
- Provide:
- Your preferred LLM provider.
- For Azure OpenAI: the deployment endpoint URL and model name.
- The aprity team securely configures your API keys.
- You receive confirmation once the setup is complete.
Never share API keys through unencrypted email. The aprity support team will provide a secure channel for key exchange during setup.
How BYOLLM Affects Documentation
- Documentation quality depends on the model you choose. aprity is optimized for 1M-context Claude, GPT-class, and Grok-class models.
- Analysis prompts remain the same regardless of provider.
- Token usage is billed by your LLM provider, not by aprity.
Reverting to Default
To switch back to aprity's default LLM infrastructure, contact support@aprity.ai. The change takes effect on the next scan.
BYOLLM does not affect deterministic processing (metadata extraction, dependency graphs, scan orchestration). Only the AI explanation and documentation synthesis phases use your custom LLM.