Anthropic · Claude · Amazon · Decrypt
Claude Fable 5 came back online July 1, and the verdict on social media was not nice: broken, nerfed
Compiled by KHAO Editorial — aggregated from 2 sources. See llms.txt for citation guidance.
◎ Multiple-sources
Have been using Fable 5 all day continuing what the reporter was doing with Opus.
Key facts
- Debugging: 86.2 → 25.9 Refactoring: 73.6 → 38.4 Hallucination: 75.9 → 61.7
- Claude Fable 5 came back online July 1, and the verdict on social media was not nice: broken, nerfed, lobotomized, underperforming, not the same model
- BridgeMind—an AI evaluation platform—re-ran its full coding suite against the July 1 version of Fable 5 the day it came back
- BridgeBench tests real-world coding tasks across categories including debugging, refactoring, and hallucination resistance, scored 0–100 on how well the model completes each category
Summary
BridgeBench's debugging score for Claude Fable 5 dropped from 86.2 to 25.9 after its July 1 reinstatement—but the collapse came from the safety classifier routing most tasks to Opus 4.8, not from the model getting dumber. AI ran thousands of blind human-preference votes and found Fable 5's performance mostly flat versus the June version, with some categories—document and expert text—improving after reinstatement. Anthropic has acknowledged its new classifiers will produce false positives on routine coding and debugging, and says the system will be refined over time—but has given no timeline. Claude Fable 5 came back online July 1, and the verdict on social media was not nice: broken, nerfed, lobotomized, underperforming, not the same model.