[FEATURE] Auto model switching based on conversation context (plan vs. implementation)
Resolved 💬 3 comments Opened Mar 5, 2026 by othmaratzmueller-bit Closed May 4, 2026
Preflight Checklist
- [x] I have searched existing requests and this feature hasn't been requested yet
- [x] This is a single feature request (not multiple features)
Problem Statement
Problem
When working on complex projects, the ideal workflow is:
- Planning/Conceptual work → cheaper/faster model (e.g. Sonnet)
- Implementation/Coding → most capable model (e.g. Opus)
Currently this requires manual /model <name> commands. It's easy to forget,
leading to either unnecessary cost (Opus for planning) or reduced code quality
(Sonnet for implementation).
Proposed Solution
Add a settings.json option for automatic context-aware model switching:
{
"autoModelSwitch": {
"planMode": "claude-sonnet-4-6",
"implementationMode": "claude-opus-4-6"
}
}
### Alternative Solutions
If automatic switching is too complex, a prompt when entering/exiting Plan
Mode would already be a huge improvement:
"You are entering Plan Mode. Switch to Sonnet to save costs? [y/N]"
### Priority
High - Significant impact on productivity
### Feature Category
Performance and speed
### Use Case Example
Example scenario:
1. I'm architecting a new feature for an enterprise AI platform (SPS TALOS)
2. I start planning: analyzing requirements, designing the approach, reviewing
existing code patterns — all conceptual work where Sonnet is sufficient
3. Once the plan is approved, I switch to implementation — now I want Opus
for maximum code quality, security awareness, and complex refactoring
4. Currently I have to manually type `/model opus` at exactly the right moment.
I forget this regularly, burning Opus tokens on pure planning, or writing
complex security-critical code with Sonnet.
With this feature:
- Enter Plan Mode → automatically switches to Sonnet (saves ~5x cost)
- Approve plan / exit Plan Mode → automatically switches back to Opus
- Zero cognitive overhead, optimal cost/quality at every step
### Additional Context
_No response_This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗