Analyst doubles down: OpenAI set to out-accelerate Anthropic by year-end

scaling01 · x · 2026-09-04

scaling01 reiterates his April call that OpenAI leading Anthropic by 1-3 months by year-end is now more likely. His reasoning: Anthropic has a better internal model but not by much, while OpenAI kept RL training for GPT-6.1-Astra; Anthropic is stuck on a model 4.16x more expensive, slower to serve and train; its reasoning RL lags (Opus 4.7 only just caught up to GPT-5.2/5.3 on math and hard science); OpenAI is just starting to hill-climb while Anthropic has had 7 months of it. Unverified speculation.

Original post →

More from Models

Models channel →