Models
Bring your own LLM credentials once, then use them in evaluations and agent workflows.
Several parts of Trodo call an LLM with your own key: LLM-judge evaluators and the AI agent and LLM nodes in workflows. You add a model credential once here, and everything that needs a model reuses it.
Model credentials are managed in the app under Integrations → Model Configurations. They are encrypted at rest, scoped to your team, and a single credential can expose more than one model.
Providers
| Provider | What you provide |
|---|---|
| OpenAI | API key. |
| Anthropic | API key. |
| Gemini (Google AI) | API key. |
| Mistral | API key. |
| Fireworks | API key. |
| Groq | API key. |
| DeepSeek | API key. |
| xAI | API key. |
| OpenAI-compatible | API key plus a base URL (for any OpenAI-style endpoint). |
| Azure OpenAI | Endpoint, API version, and deployment name. |
| Google Vertex AI | Project, location, and a service-account key (or API key). |
| AWS Bedrock | Access key, secret key, and region. |
Add a model
Open Model Configurations. In the app, go to Integrations and open Model Configurations.
Pick a provider and enter its credentials from the table above.
Add one or more models to that credential (for example gpt-4o and gpt-4o-mini on one OpenAI key).
Use it. The model is now selectable wherever Trodo calls an LLM: an LLM-judge evaluator or an AI agent node.
Where models are used
- Evaluations: an LLM judge scores runs with a model you add here (OpenAI, Anthropic, or Gemini).
- Agents: the AI agent and LLM nodes run on a model you add here, across all the providers above.