Equity research test finds GPT-6 Astra behind Fable 5.1

Paradis @ParadisLabsOriginally posted 2026.09.07

Open original on X

After testing GPT-6 Astra all weekend, Fable 5.1 is far superior (for me) on traditional equity research tasks—more coherent lines of reasoning, fewer material oversights, and more unique and interesting insights that warrant additional token usage. GPT-6 Astra can certainly be a good start point for equity research. But I still think Fable 5.1 is more of a one-stop-shop model. Plenty of room for improvement though—for both models—which is the exciting bit going forward!