Skip to content
← Models
Open weightsAllen Institute for AI/10 trustVerified

AstaBrief-8B

Ai2's open 8B model that writes cited science reports in one forward pass.

Visit official page →

About

AstaBrief-8B is an open-weights 8-billion-parameter model from the Allen Institute for AI, released October 2, 2026 alongside its weights and training data. Built on Qwen3-8B with supervised fine-tuning and direct preference optimization, it takes a research question plus retrieved literature excerpts and writes a fully cited scientific report in a single forward pass.

It powers Fast mode in Asta, Ai2's agentic platform for scientific work, next to the Claude-powered Thinking mode. Ai2 reports the full pipeline averages 51.1 seconds per report in Fast mode versus 178.5 seconds in Thinking mode, roughly 3.5 times faster, while matching report quality on its internal evaluations.

The model is licensed under Apache 2.0 and is small enough to run on a single GPU, which lets institutions generate cited reports behind their own firewall when research questions touch sensitive or unpublished work. Ai2's speed and quality figures are self-reported; the 47,000 filtered training reports are released for independent scrutiny.

Specs

Context window

Pricing

Free quota

API access

Weights

Open weights

Released

October 2, 2026

Benchmarks

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

scientific-writingliterature-reviewresearch-assistants

Trust score — 8/10

/10 trustVerified

Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 104; Trending score 78/100 from researcher assessment.

Sources