Science
OpenAI had Not resumed training on its models
Anthropic had Not launched a new Claude Fable model, and a different AI system led the performance leaderboard on Oct 12.
OpenAI had Not resumed training on its models as of Wednesday, 14 October. At the same time Anthropic had Not launched a new Claude Fable model, and a different system occupied the top spot on the performance leaderboard on 12 October.
OpenAI's training pause continued beyond its previously anticipated restart window. The company kept its large-scale model training halted through at least 14 October, despite industry speculation of a later-month resumption. No new version of its flagship model was released during that period.
Anthropic's schedule for the Claude Fable series also lagged. The firm did Not issue a new Claude Fable model before 12 October, contrary to expectations of an early-October launch. Existing Claude Fable versions remained the latest publicly available offerings.
Performance benchmarks published on 12 October listed a model other than Claude Opus 5.5-High as the top performer. The leading system outperformed the previously dominant Claude Opus variant across the evaluated metrics.
Analysis
Competitors could leverage the gap to attract research talent and commercial interest, while OpenAI risked falling behind in benchmark scores and user adoption. The pause also signaled internal constraints, whether technical, regulatory, or strategic, that limited immediate progress.
Anthropic's failure to release a new Claude Fable model before 12 October delayed the expected expansion of its product line. Without an updated model, Anthropic missed an opportunity to showcase improvements that might have appealed to enterprise customers seeking the latest capabilities. The delay may affect its positioning against rivals that are actively rolling out newer versions.
The emergence of a different model as the top performer on 12 October demonstrated that leadership in AI performance was Not static. This shift could influence research funding, partnership choices, and the direction of future development as firms chase the newly established benchmark. However, the lack of detail about the leading model leaves uncertainty about which organization gained the advantage and how durable the lead will be.
What this was built from
Each claim above traces to a market below, at the price it was trading when this edition went to press.