What’s New?
This release introduces several major features that enhance the Clarity experience for end users:
- Model-Agnostic Agents with Agent Parity: The new Model Agnostic Agents functionality enables agents and tools to use different models (GPT, Claude, Gemini) best suited to each task. These new agents have been built with agent parity in mind, ensuring consistent behavior across models. Capabilities not natively supported by one model are now enabled consistently for all.
- Enhanced Code Interpreter Tool: The code interpreter has been significantly improved with structured data handling (automatic previews of pandas DataFrames), a protocol that guides LLM code generation to produce results of interest, and improved error handling that provides clear feedback to users (e.g., when images exceed available memory in the sandbox).
- Multiple File Uploads: You can now upload up to 10 documents at a time with a single request, making it faster and more convenient to share multiple files with your agent. You can then ask the agent to tell you what files are available in your conversation where they were uploaded.
- Artifacts Improvements: Improved artifact display and organization for better visibility into tool execution details make it easier to understand how the agent processed your request. For example, the code interpreter tool’s “Code” artifact now makes it easier to interpret what the code tool generates and executes.