Reproduce:
- Send a message in the chat and wait for completion.
- Edit that message, change mode or model you’d like to use for this message and send (continue without reverting)
- The original mode and model will still be used for the request
Besides wasting a few credits, this blocks a powerful workflow, where you can continue working with the exact chat context you need for the next task (aka branching). Would love to see this fixed soon.
Edit: If the team is considering actual conversation branching at some points soon, that would be even more powerful!