What LLM providers does Agenta support?

Agenta has built-in support for OpenAI, Anthropic, Google Gemini, Mistral AI, Cohere, Groq, Together AI, OpenRouter, Perplexity AI, DeepInfra, and MiniMax. For these providers you add your API key under Settings > Model Hub and pick from a maintained model list.

Agenta also supports Azure OpenAI, AWS Bedrock, and Google Vertex AI. These are added as custom providers because they need more than an API key (for example a deployment URL, an AWS region, or service account credentials).

You can add any OpenAI-compatible endpoint as a custom provider. This covers self-hosted models (for example through Ollama), fine-tuned models, and custom deployments.

You can also add models that are missing from a built-in provider's list, for example a newly released model on OpenRouter.

See custom providers for setup instructions.