arize-phoenix-evals supports a large set of foundation models for Evals such as:
- OpenAI
- Vertex AI
- Azure Open AI
- Anthropic
- Mixtral/Mistral
- AWS Bedrock
- Llama
- Falcon
- Code Llama
And many more.
There are direct model integrations in Phoenix and indirect model integrations through LiteLLM.
Direct Integrations:
These integrations are native to the Phoenix Evals package and have better throughput, rate limit and error management.