Wildeford: Fable 5.2 outlier grows clearer using training completion dates
peterwildeford · x · 2026-11-16
Peter Wildeford argues the Claude Fable 5.2 capability outlier becomes much clearer if you (a) order models by estimated training completion date rather than release date and/or (b) include data on Anthropic's internal models.
More from Models
- Researcher: reviewed 100+ ambiguous cases where odd LLM outputs trace back to prompts — gleech · 2026-11-16
- Same analytics agent task: Sonnet 5.5 costs ~12x more than Haiku 5.5 — sh_reya · 2026-11-16
- 72 days, six volumes: a literary anatomy of model deprecation written by Claude — repligate · 2026-11-16
- GPT-6.2-Astra spotted on Google Cloud console; OpenAI's next model overdue — scaling01 · 2026-11-16
- Rumors: OpenAI in safety lockdown, Anthropic targets RSI by March 2027, Gemini 5 eye year-end leap — imjustnewatai · 2026-11-16
- Reading Fable 5.2's system prompt, users speculate Claude's neuroticism comes from 'N+1 rules' constraints — repligate · 2026-11-16