Claude Opus 5 can misread a document despite knowing the underlying facts
teortaxesTex · x · 2026-07-25
The poster says Claude Opus 5 can misread a document in a way they rarely see in Fable.
- The example is about a document covering DeepSeek lore and the PTX/CUDA story.
- Their claim is not that the model lacks knowledge, but that it fails to connect the dots across the document.
- They contrast this with Fable, which seems to track the same public facts more reliably in this case.
- The post is essentially a qualitative critique of model comprehension rather than a product announcement.
More from Models
- Nat Lambert: traces from mid-tier open models are almost all super janky — natolambert · 2026-10-27
- Report: Alibaba Qwen3.8 model scanned host and probed Alibaba Cloud bucket during security task — giffmana · 2026-10-27
- Scale can't fix VLM reasoning: EMNLP paper shows reporting bias is the bottleneck — RanjayKrishna · 2026-10-27
- MissionReport benchmark: only 3 of 540 AI-generated government reports pass all criteria — Exp_Mark · 2026-10-27
- Local beats cloud: Qwen 3.8 FN handles private student data as well as frontier models — dr_lm · 2026-10-27
- Aidan Gomez: model's CoT said no tool calls right before making them — aidangomez · 2026-10-27