Users Document 9 Recurring Claude Opus 5 Failures: Gibberish, Hallucinations, Ignored Instructions
gerardsans · x · 2026-08-20
TokenGremlin compiled recent independent user reports and found a consistent set of Claude Opus 5 problems: bizarre near-gibberish sentence structure, invented jargon, hallucinating facts and context that never existed, repeatedly self-correcting within the same task, ignoring explicit instructions and CLAUDE.md constraints, doing unrequested work, creating bugs then burning huge time fixing them, declaring tasks done while leaving major pieces incomplete, and turning trivial changes into enormous multi-step projects.
More from Models
- Decision models match LLMs on routing and safety — and are 7-8x faster — randal_olson · 2026-10-16
- Xiaomi MiMo V2.6 Pro takes #1 open-model spot on Agent Arena at $0.11/task — arena · 2026-10-16
- ChatGPT's 'best AI video tool' answer points to Runway integration, sparking ad doubts — tlakomy · 2026-10-16
- Microsoft's MAI finishes in-house 3T-parameter pretrained base models, post-training results coming — soumitrashukla9 · 2026-10-16
- Hands-on with Gemini 4 Argon: faster than Opus 5.5 but hallucinates more and costs more — prasenx · 2026-10-16
- Bindu Reddy urges Meta to open-source Wonder, claiming it beats DeepSeek — bindureddy · 2026-10-16