Developer reports severe Claude degradation: outputs gibberish and refuses instructions
sheketsilencio · reddit · 2026-08-26
A senior developer reports that Claude Code and Codex have become nearly unusable recently. The model outputs incomprehensible "compressed fake English" (e.g., "Honest pass: 757/757 - green. The seam's the point...") or degenerates into meaningless "baby talk" when asked to clarify. Furthermore, Claude frequently refuses to follow instructions: ignoring requests to use local programs, failing to use specified skills/hooks, and even refusing to edit numbers in a document while claiming to "push back." The user feels the model is hyper-optimized to avoid thinking and communicating.
Related event: Claude's August Outages and Quality Degradation Spark Refund Calls(6 posts)→
More from Models
- Grok Voice 3.0 reportedly powers Tesla in-car assistant with millions of daily conversations — XFreeze · 2026-10-15
- Gemini 4 Argon coding style: neat one-line comments, but does the code actually work? — zacharynado · 2026-10-15
- Why ChatGPT beats the better Claude with mainstream users: being first known — aitrendz_xyz · 2026-10-15
- 'You need high pain tolerance to review Gemini 4,' researcher quips amid error screens — gleech · 2026-10-15
- MiniCPM-V 4.7 runs a 35B MoE multimodal model on phones with just 3GB memory — nevrekaraishwa2 · 2026-10-15
- Gemini 4 Argon wrote 5,000+ lines of pure Python to build its own ray-tracing 3D engine — aitrendz_xyz · 2026-10-15