Evaluation Models

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

• Falcon

• Code Llama

• Llama3

• Deepseek

• Deberta

• DBRX

• Qwen

And many more.

There are direct model integrations in Phoenix and indirect model integrations (e.x. local modals) through LiteLLM.

Direct Integrations:

These integrations are native to the Phoenix Evals package and have better throughput, rate limit and error management.

Vertex AI

OpenAI

Azure OpenAI

Anthropic

Mistral