📊Hebbia's GPT-6 Astra Beats Rivals by 13.6 Points on Financial Research
Your banker's deck just got a lot more trustworthy
TL;DR
Hebbia's prerelease tests show GPT-6 Astra citing sources with 99.5% relevance and hitting 97.9% of presentation requirements. Finance teams building decks from messy data rooms should pay attention, since the sources stay traceable back to the original cells.
Hebbia says GPT-6 Astra beat every alternative it tested on financial research, and by a wide margin. It led by 13.6 percentage points on citing the required source types and by 14.4 points on meeting presentation requirements. If you build tools for bankers or analysts, this one matters. The model doesn't just answer questions. It selects the documents or Matrix cells behind each answer and turns the analysis into native PowerPoint slides that follow the user's structure. That means less time reformatting decks and checking figures by hand. The test setup was specific: a prerelease build run in August and September 2026, using identical financial cases and grading criteria, with domain experts auditing the rubric against source documents. Cited excerpts were 99.5% relevant, and 97.9% of applicable presentation requirements were met. Hebbia's pitch is traceability, which is the part worth watching as these models move into real deal workflows.

Key Points
Cited excerpts from GPT-6 Astra were relevant to the question 99.5% of the time in Hebbia's internal financial research tests run in August and September 2026.
Keep reading — it's free
Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.
Already a member? Sign in
Comments
Be the first to comment