Feature Request: LLM Flex Switch — Seamless Model Swapping in Chat
Summary:
Allow users to toggle between GPT models (e.g., GPT-4o, GPT-4, GPT-3.5) within an ongoing conversation, without losing chat history or starting a new thread.
Why This Matters:
Currently, switching models requires starting a new conversation and losing context. This adds friction and limits experimentation. A seamless in-chat model toggle would unlock:
- Efficiency: Quickly rerun tasks in GPT-3.5 for speed/cost
- Power: Deep dive with GPT-4o when needed for complex logic, visuals, or file interpretation
- Comparisons: A/B test creative output, code, or tone between models
- Control: Users regain flexibility depending on task priority (speed, quality, or cost)
Suggested Implementation:
- A dropdown toggle at the top of the chat, labeled:
makefile
CopyEdit
Model: GPT-4o ⏷
- Optionally auto-suggest model switches for performance-sensitive tasks.
- Messages could include an indicator of which model generated them.
Bonus Ideas:
- Tag specific prompts to rerun in a different model on demand.
- Visual badge (e.g., GPT-4o or
GPT-3.5) beside each response. - Hybrid Mode: Let users set a chat to “default to 3.5 unless advanced reasoning is detected.”
User Feedback (me):
As a power user juggling sales, technical documentation, AI consulting, and design, I often need both the speed of GPT-3.5 and the advanced capabilities of GPT-4o. A live switch would massively boost productivity and reduce repetitive context-setting