From the source
- Hebbia tested
OpenAI's GPT-6 Astra on financial research tasks and reported that it outperformed the best alternative it tested on citing required source types and on meeting presentation requirements.
- The company, whose Matrix product links answers to the documents behind them, ran the tests on a prerelease build in August-September 2026 using the same financial research cases and grading criteria for each model, with domain experts writing or auditing the criteria against source documents.
- GPT-6 Astra scored 13.6 percentage points above the best alternative tested on citing the required types of sources, and 99.5% of the excerpts it cited were relevant to the question.
- On presentation, it met 97.9% of the applicable requirements, a 14.4 percentage-point lead over the best alternative tested, and ranked first on both following the brief and presentation quality.
- Hebbia says other models traded off polish and adherence to the brief, while GPT-6 Astra was the first it has seen to rank first on both.
- In Matrix, users can open a citation to see the underlying document and highlighted passage, and slide footnotes lead back to the source material.
- Hebbia's Kavita says the model's precise outputs could cut the time bankers spend reformatting decks and checking sources, and describes a future role for agents that monitor information in the background and surface signals.






