We have two new model choice options in OpenHands/software-agent-sdk and OpenHands/agent-canvas:
- the /model command that allows for switch models manually
- the model selection tool that allows the agent to dynamically switch between models
We should create documents, placed in the appropriate place, that document this functionality. Both should have instructions accompanied by screenshots within the agent canvas.
For the both examples, it would be good to show with three profiles: claude-opus-4-7 (a more expensive model that excels at frontend design), gpt-5.5 (a model that excels at instruction following), and minimax-m2.7 (a much less costly model that can be used as a daily driver).
The /model command can demonstrate the user using the gpt-5.5 for planning, and then having the user switch to minimax for the actual implementation.
The model selection tool can demonstrate the following prompt:
"Implement a simple web page on the features of OpenHands with Claude Opus 4.7, and then switch to GPT-5.5 and review the code."
We have two new model choice options in OpenHands/software-agent-sdk and OpenHands/agent-canvas:
We should create documents, placed in the appropriate place, that document this functionality. Both should have instructions accompanied by screenshots within the agent canvas.
For the both examples, it would be good to show with three profiles: claude-opus-4-7 (a more expensive model that excels at frontend design), gpt-5.5 (a model that excels at instruction following), and minimax-m2.7 (a much less costly model that can be used as a daily driver).
The /model command can demonstrate the user using the gpt-5.5 for planning, and then having the user switch to minimax for the actual implementation.
The model selection tool can demonstrate the following prompt:
"Implement a simple web page on the features of OpenHands with Claude Opus 4.7, and then switch to GPT-5.5 and review the code."