Hebbia cites GPT-6 Astra gains in financial research tests
Original title: Hebbia sees improvements in financial research with GPT-6 Astra
Hebbia’s tests of a prerelease build put citation quality 13.6 points and presentation compliance 14.4 points above the next model.
OpenAI published a customer story in which financial-research firm Hebbia says GPT-6 Astra improved source selection and presentation building in its Matrix product. In Hebbia’s grading, the model cited required source types 13.6 percentage points above the best alternative, and 99.5% of cited excerpts were relevant. It met 97.9% of applicable presentation requirements, 14.4 points ahead of the next model, and ranked first on both brief-following and slide quality.
Unconfirmed: The scores are Hebbia’s own tests of a prerelease build, not an independent benchmark.
- Hebbia tested a prerelease build in August and September 2026.
- The same financial cases and grading criteria were used for every model.
- Domain experts wrote or audited criteria against source documents.
- A hypothetical workflow in the story involves about 2,000 data-room files.
- Products
- MatrixPowerPoint
- AI models
- GPT-6 Astra
From the original article
3 min read at openai.com
Hebbia tested GPT-6 Astra for financial research. Its prerelease evaluations found gains in source coverage, citation relevance, and presentation quality.
openai.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.