Reasoning Models
Reasoning models represent a significant advancement in how AI agents handle complex, multi-step tasks. When combined with Sema4's agent tuning, these models deliver dramatically improved accuracy and reliability for enterprise workflows.
Sema4.AI agent accuracy with τ2-telecom benchmark
Supported Providers and Models
| Provider | Recommended Reasoning Models |
|---|---|
| OpenAI, Microsoft Azure OpenAI | OpenAI GPT-5 Low, GPT-5.1 Codex Max Medium |
| AWS Bedrock, Snowflake Cortex | Claude Sonnet 4.5 Medium |
To use reasoning models (GPT-5, Claude 4 and above), you'll need version 1.6 or later across all Sema4.AI products.
For a full list of supported models, see the Model Compatibility & Benchmark page.
Getting Started
- Setup your model provider — Configure credentials for Azure AI Foundry, AWS Bedrock, Google Vertex AI, or Gemini API
- Select a reasoning model — When creating or editing an agent, choose your model and reasoning level (Low/Medium/High) from the Large Language Model dropdown
- Troubleshoot issues — Common problems and solutions when working with reasoning models