Claude Fable 5 entered the published record on June 9, 2026.

Epoch AI's catalog identifies Claude Fable 5 as an Anthropic model released June 9, 2026. It records API access, the domains Language, Multimodal, and the tasks Language modeling/generation and Question answering.

Those fields establish identity, timing, access, and declared capability categories. They do not measure adoption, revenue, inference volume, or market share.

In LMArena's July 12, 2026 text style-control overall snapshot, Claude Fable 5 is the provider-frontier entry for Anthropic with a 1507.5 point estimate and a 95% confidence interval from 1500.0 to 1515.0.

The row is based on 7,856 battles. It describes performance in that table, not overall product superiority.

The defensible statement is narrow: a published launch record precedes a leading provider-frontier result in a later published snapshot.

The artifacts do not support claims about causal lift, user migration, revenue, developer adoption, or performance before and after launch. Those would require separate time-series and market evidence.

• Report the launch date and source.
• Report the later standing with its confidence interval and battle count.
• Keep sequence separate from causation.
• Revisit the account when comparable later snapshots pass review.

A public preference result can justify evaluation priority. It cannot finish procurement.

Teams should test representative work, adversarial cases, structured-output reliability, latency, full-system cost, rate limits, security terms, regional access, and model-change policy before deployment.

1. This report uses one catalog record and one point-in-time LMArena snapshot.
2. It does not measure causal launch impact, adoption, revenue, usage, price, latency, safety, or reliability.
3. The July 2026 point-in-time snapshot is not joined to the legacy archive.
4. No Model Gauntlet composite score is calculated.

See the methodology and corrections policy.