User Test: Opus 5 Suffers from Severe Long-Context Failure

ATTlKA · x · 2026-08-24

A user criticizes the current performance of Anthropic's Opus 5 model as abysmal, citing a severe lack of long-horizon coherence. The model reportedly descends into incoherence after just two or three turns of back-and-forth conversation, failing to maintain context over extended interactions.

Original post →

More from Models

Models channel →