Anthropic's latest AI model, Claude Opus 5, has achieved a remarkable 30.2 percent score on the ARC-AGI-3 benchmark, significantly outperforming the previous record set by GPT-5.6 Sol at just 7.8 percent, according to The Decoder.
The developers of the ARC-AGI-3 benchmark noted that Claude Opus 5 demonstrated an unprecedented capability by independently formulating reflection equations, a behavior not observed in any other model so far. This breakthrough highlights the rapid advancements in AI reasoning and problem-solving skills.
For Japanese investors and tech markets, progress like this signals accelerating innovation in AI technologies, which could influence developments across FX algorithmic trading, crypto analytics, and equities market forecasting.