Notably, Claude Opus 4.1 and 4—the most recently released and most capable models of those that we test—perform the best in our experiments, suggesting that introspective capabilities may emerge alongside other improvements to language models.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Claude Opus 4 and 4.1 exhibit the greatest degree of introspective awareness among tested modelsclaim0.887
  Based on consistent best performance across experiments.
• Our results demonstrate that modern language models possess at least a limited, functional form of introspective awareness.quote0.844
  Abstract's main conclusion.
• Introspective capabilities may continue to develop with further improvements to model capabilitiesclaim0.832
  Forward-looking statement about future models.
• Modern language models possess at least a limited, functional form of introspective awarenessclaim0.825
  The paper's central interpretive assertion.
• We hypothesize that introspective capabilities may scale with model size and architecture, including recurrence/looping that extends the integration windowhypothesis0.816
  Forward-looking prediction about whether early-layer introspection generalizes to larger models or recurrent architectures
• Emergent Introspective Awareness in Large Language Models (Lindsey, 2025)concept0.815
  Related work demonstrating LLM introspective capabilities with scale-dependent pattern paralleling ESR
• In Opus 4.1, representation of the think word decays to baseline by the final layer, unlike Claude 3 models where it persistsfinding0.812
  Suggests that later models can keep the thought 'silent' rather than letting it influence output.
• Introspective capabilities have threshold effects requiring very large models; 70B models are barely on the threshold, and independent researchers lack access to larger models.claim0.808
  Practical bottleneck explaining why these phenomena are not widely studied.