Updated

AstaBrief 8B

Ai2 released AstaBrief 8B, an open-weights model that turns a research question and retrieved literature into a cited scientific report in one pass. The model and training data are available for download; it is also offered as Fast mode in Ai2’s Asta platform.

By No Code MBA · Facts checked

Provider
Ai2
Announcement date
Status at last check
Generally available
Model family
Asta
Capabilities
Text

What changed

AstaBrief 8B is an open model fine-tuned from Qwen3-8B that turns a research question and retrieved literature excerpts into a cited report in one pass. Ai2 trained it with about 47K SFT examples and 6K DPO preference pairs drawn from Asta queries. Ai2 reports an average of 51.1 seconds per report in Asta, about 3.5x faster than the Claude-powered Thinking mode, and says it was competitive with that pipeline and DR Tulu on several answer and citation measures.

Access and availability

Download weights from Hugging Face (allenai/AstaBrief_8B) under Apache 2.0, with training data in the allenai/astabrief collection. It also runs as Fast mode in the Generate a report feature at asta.allen.ai.

Official access documentation

What you can build

Try it on a small set of research questions with known source papers, then check citation support, omitted caveats, and whether the report is useful as a first draft.

Limitations

Ai2 says training and evaluation were mostly done in 2025 and were not rerun against current frontier models, and the human study was small (14 questions, three researchers). It needs the recommended prompt format and retrieved passages, and was trained on sequences up to 16K tokens.

Sources and release history