Hebbia sees improvements in financial research with GPT-6 Astra
, updated First seen Updated Limited coverage, tapering
OpenAI · Models
Reported by OpenAI1 source ↓
Hebbia’s prerelease tests found GPT-6 Astra ahead of the best alternative on source coverage and presentation requirements; 99.5% of its cited excerpts were relevant to the question.
Key points
- It scored 13.6 percentage points above the best alternative on citing required source types
- It met 97.9% of applicable presentation requirements, 14.4 percentage points above the best alternative
- Hebbia tested a prerelease build from August to September 2026
Details
Hebbia tested the model on financial research cases using the same cases and grading criteria for each model. Domain experts wrote or audited the criteria against source documents.
In Hebbia’s Matrix product, the model selects documents or cells to support answers and can turn analysis into PowerPoint slides. Users can open citations to view the underlying document and highlighted passage; slide footnotes link back to source material.
Written from coverage as of
Images
- OpenAI
- OpenAI
- OpenAI
- OpenAI