Skip to content
important New models

AstaBrief 8B model released for scientific reports with citations, weights and training data also available

clearly official source

The AstaBrief 8B model generates scientific reports with citations and is available in the Fast mode of the Asta service. The authors are releasing both the weights and the training data. According to their measurements, the entire process takes an average of 51.1 seconds compared to 178.5 seconds in Thinking mode.

The authors have released the AstaBrief 8B model, which generates a report with citations from a research question and retrieved excerpts of scientific literature. It is available in the Generate a report feature of the Asta service as Fast mode alongside the Thinking mode, which uses Claude models. At the same time, the authors are releasing the model's weights and training data. It is based on the Qwen3-8B model, modified using supervised fine-tuning (SFT) and direct preference optimization (DPO).

The model writes the entire report in a single pass and omits some intermediate steps used in Thinking mode. According to the authors, the entire process in Fast mode takes an average of 51.1 seconds compared to 178.5 seconds in Thinking mode, i.e., roughly 3.5 times shorter. Most of the training and evaluation took place in 2025; the authors did not repeat a full evaluation against current state-of-the-art models.

Open weights allow institutions to run the model on their own infrastructure, which the authors link to the need to protect sensitive or unpublished research questions. The release also includes an example procedure for generating reports from your own PDFs as a basis for local deployment. See the source article for details.

What changed

Why it matters

In Fast mode, researchers can obtain a preliminary report with citations more quickly and then develop it further; the stated speedup is based on the authors' measurements. Institutions gain the ability to set up their own operation for working with sensitive research questions and their own PDFs.

Release card

AstaBrief 8B

Ai2

open weights
Inputs
The input is a research question and retrieved text excerpts from the scientific literature. The output is a report with citations.
Availability
The model is available in the Fast mode of the Generate a report feature of the Asta service. The weights and training data are publicly released; the model can be downloaded and run on your own infrastructure.
Documented measurements
According to the sources, it is suitable for
  • Generating scientific reports with citations from retrieved literature.
  • Local generation of reports from your own PDFs using the published example procedure.
  • Operation on your own infrastructure for sensitive or unpublished research questions.

The authors compare the model to proprietary models used for scientific reports and to the Thinking mode of the Asta service, which uses Claude models. They did not repeat a full evaluation against current state-of-the-art models.

The card summarizes information from the article and any dated corrections, with a link to the original source. It is not our assessment of the model. It does not yet have a dedicated editorial profile. Model selection and other announcements →

Two audiences, two different impacts

What this means

01

For individuals

For research, you can use the Fast mode of the Asta service to get a preliminary report with citations more quickly, which can then be further developed.

What to do Try the Fast mode of the Asta service on a research question and verify the conclusions against the cited sources.
More practical updates →
02

For a business

Thanks to open weights, institutions can prepare operation on their own infrastructure for research involving sensitive or unpublished research.

Risks and compliance
What to decide Assess the published procedure for generating reports from custom PDFs as a basis for internal deployment.
More business impacts →
Asta AstaBrief 8B

Check the original

Event sources

clearly official source · 1 vydavatel, 0 independent. We count feeds from the same owner only once.

1
Hugging Face Blog primary source · first detected Open-sourcing AstaBrief, the fast report-generation model in Asta