Performance Benchmarks and Market Activity
Anthropic’s latest model, Claude Fable 5.2, has reportedly taken the lead on SimpleBench, a benchmark designed to measure cognitive reasoning. According to data discussed on Reddit, Fable 5.2 achieved a score of 94.9%, surpassing Google’s Gemini 4 Argon, which scored 91.7%, and Gemini 4 Carbon, which recorded 92.6%.
The performance metrics suggest a significant shift in the competitive landscape of large language models. Unverified discussions on social media platforms have noted that these scores now exceed the average human baseline. While some users on Reddit have suggested that the average human score sits at approximately 83.7%, others have argued that the gap between AI performance and human cognitive ability is closing rapidly. Specifically, commenters noted that while the best human performers may still hold a slight advantage in certain nuances, the current trajectory suggests AI is approaching, or has already surpassed, human-level performance in tasks involving word games and spatial reasoning.
Anthropic Market Developments
In parallel with these technical advancements, Anthropic has seen significant activity in the public markets following its Initial Public Offering (IPO). According to reports from Reddit, Anthropic shares opened for trading at $150. As of the most recent market updates cited in these discussions, the share price has seen a slight decline, trading at approximately $143.
Technical Observations and Robotics
Beyond linguistic and reasoning benchmarks, observers have begun evaluating Fable 5.2’s capabilities in specialized domains. There has been specific interest in the model’s performance regarding robotics benchmarks, though detailed, independent verification of these results remains pending.
Some users have also commented on the architectural efficiency of the new release. In comparisons between Fable 5.2 and previous iterations, such as Fable 5.1, some observers have noted differences in model scale and “needless mass,” suggesting a shift in how Anthropic is optimizing its model architecture for specific tasks.
