[Feature Request] Investigate and optimize token consumption for lower-tier models on Max Plan

Bug Description
I have the Claude AI Max Plan. I'm using the Sonora model. My tokens are being used up insanely fast. If this continues I will need to switch plans. When is a fix coming? This is ridiculous.

Environment Info

• Platform: win32
• Terminal: vscode
• Version: 2.1.86
• Feedback ID: 7ca17b3f-3175-4959-9d69-587ec574d4f5

Errors

[{"error":"Error: Request was aborted.\n    at mk8 (B:/~BUN/root/src/entrypoints/cli.js:1421:16708)\n    at next (native:1:11)\n    at $h_ (B:/~BUN/root/src/entrypoints/cli.js:7618:8934)\n    at next (native:1:11)\n    at _h_ (B:/~BUN/root/src/entrypoints/cli.js:7623:10737)\n    at processTicksAndRejections (native:7:39)","timestamp":"2026-04-03T13:48:47.640Z"}]