PEGASUS
Abstractive summarization model pretrained with gap-sentence generation.
About
PEGASUS was proposed in December 2019 by researchers at Google Brain and Imperial College London. Its pretraining objective, gap-sentence generation, masks out important sentences from a document and trains the model to generate them together, making pretraining closely resemble the summarization task.
The 568-million-parameter model achieved state-of-the-art ROUGE scores on all 12 downstream summarization tasks it was evaluated on, spanning news, science, stories, patents, and legislation, and showed surprisingly strong low-resource performance with only 1,000 examples. It is released under Apache 2.0.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 82; Trending score 62/100 from researcher assessment.