OpenAI’s latest language model, GPT-6 Astra, has delivered mixed benchmark results. According to Epoch AI, GPT-6 Astra scored 169 points, leading current benchmarks, while Artificial Analysis reported that the model performed no better than its predecessor and lagged behind competitor Claude Fable 5.1.
Despite these mixed outcomes, GPT-6 Astra demonstrated a significant breakthrough on the ARC-AGI-3 test, working more efficiently than the average human for the first time, as noted by The Decoder. This milestone has led François Chollet to accelerate his forecast for the arrival of artificial general intelligence (AGI), stating that progress is running twice as fast as expected.
For Japanese investors and tech markets, this development underscores the rapid pace of AI innovation, which could influence future FX and equities trading strategies as AI-driven technologies continue to evolve.