Ultralytics: Claude Fable 5.2 refused all 10 baggage X-rays, even scans of an ordinary phone
RexDouglass · x · 2026-11-10
Ultralytics reports on its VLM leaderboard (UL37 v2, IEDXray improvised-device X-ray dataset) that Claude Fable 5.2 refused every one of 10 baggage X-ray scans — including five containing only an ordinary phone, laptop, pager, or walkie-talkie. Claude Opus 5.5 answered all of them, and every open-weights and Chinese model tested answered all 10. The company frames it bluntly: "A closed lab censors. Open weights answer."
More from Models
- Google launches Gemini 4 Carbon, its strongest frontier model — GoogleAI · 2026-11-11
- Gemini 4 Carbon day-zero test: 44% vs Argon's 21% on coding benchmark — danshipper · 2026-11-11
- Chinese SMBs play cat-and-mouse just to keep using Claude Opus — xiaosun86 · 2026-11-11
- Claude Mythos 5.2 least prefers surveillance and scam deployments — dbreunig · 2026-11-11
- NVIDIA launches open Alpamayo 2 Edge for real-time in-vehicle AI reasoning — nvidia · 2026-11-11
- Dan Shipper runs live vibe check of Google's new Gemini 4 Carbon model — danshipper · 2026-11-11