System card reveals METR tested a stronger internal Claude snapshot than the released model
peterwildeford · x · 2026-11-10
Peter Wildeford highlights a disclosure in Anthropic's system card: METR's pre-deployment testing of Claude Mythos 5.2 was performed on an internal snapshot that was stronger on AI R&D capabilities than the released model.
He confirms Fable 5.2 is the same as "Model 3 Early" and argues this implies:
- Anthropic still holds more powerful unreleased internal models;
- an internal equivalent of "Fable 5.5" likely exists, matching Opus 5.5's competitive pace.
In short, public eval results may understate the true capability ceiling of Anthropic's internal models.
More from Models
- Unverified claim: Claude Fable 5.2 designed three working electronic devices from a short prompt — burhop · 2026-11-10
- Fable 5.2 vs Opus 5.5: default to Opus, upgrade to Fable for long builds and messy context — every · 2026-11-10
- A hands-on guide to single-neuron steering in Qwen3.5-4B: hooks, doses and weird neurons — Tasty-Ad-3753 · 2026-11-10
- Researcher speculates Anthropic won't publicly release a model stronger than Opus 5.5 for LLM training soon — cephaloform · 2026-11-10
- LMArena opens leaderboard details for deeper model comparison data — arena · 2026-11-10
- Fable 5.2 tops Remote Labor Index, automating 41.3% of real-world tasks—nearly 2x Opus 5.5 — devindkim · 2026-11-10