--- title: README emoji: 📚 colorFrom: blue colorTo: green sdk: static pinned: false --- # Glasp Glasp is a web and PDF highlighter used by more than 3.5 million people. They use it to read and annotate research papers, articles, PDFs and videos, so the community leans toward researchers, students and working professionals. - Tens of thousands of people studying medicine and the health sciences use Glasp. - More than 40 universities and colleges list or recommend it. - About two in three people who highlight research papers on Glasp use it in English. ## Work with Glasp Members can opt in to hear about paid AI projects. We invite the ones who match your brief; screening, contracting and payment can stay in your own process. Open for requests: [Medical review by medical students](https://huggingface.co/datasets/glasp/medical-student-review) · [Expert evaluation](https://huggingface.co/datasets/glasp/expert-evaluation) · [PhD-level research tasks](https://huggingface.co/datasets/glasp/phd-research-tasks) **AI data buyers and data teams** Experts for evaluation and annotation, speech and video recorded to your spec, native speakers in many languages. Contact: [hi@glasp.co](mailto:hi@glasp.co?subject=HF%3A%20data%20request) **AI labs** Researchers who run your agents through long research tasks, find where models fail, and write and grade evaluation tasks. Contact: [hi@glasp.co](mailto:hi@glasp.co?subject=HF%3A%20lab%20request) **Universities and research groups** Participants and expert annotators for your benchmark or study. Glasp charges no fee for academic research; how participants are paid follows your study's own terms. Contact: [hi@glasp.co](mailto:hi@glasp.co?subject=HF%3A%20research%20request) ## Who reads with Glasp When people join Glasp they can tell us their field and their education. The shares below are from members who joined between October 14 and November 19, 2026 and answered those questions. Answers are self-reported and rounded. This is a sample of recent members, not the whole user base. About one in five hold a Master's degree or higher. | Highest education | Share | |---|---| | PhD | 3% | | Professional degree (for example medicine or law) | 3% | | Master's | 12% | | Bachelor's | 31% | | Some university | 15% | | Secondary | 19% | | Prefer not to say | 17% | | Field of study or work | Share | |---|---| | Software | 17% | | Business | 12% | | Arts and media | 8% | | Education | 8% | | Data and AI | 7% | | Finance and economics | 5% | | Engineering | 5% | | Medicine and health | 4% | | Social sciences | 3% | | Law | 3% | | Languages | 3% | | Mathematics | 2% | | Humanities | 2% | | Life sciences | 2% | | Physics | 1% | | Chemistry | 1% | | Other or undeclared | 16% | Members who said they would like to hear about AI research projects speak English, Spanish, Vietnamese, Portuguese, Korean, Russian, Chinese, Arabic, French, Japanese, Indonesian and Hindi. ## Research from the Glasp team - [Glasp Research](https://glasp.co/research): research published by the Glasp team. - [Glasp Research on Substack](https://read.glasp.co/t/glasp-research): articles by the Glasp team on AI data, expert work and research. - [YouTube ASR benchmark against human captions](https://huggingface.co/datasets/glasp/youtube-asr-human-caption-benchmark): error rates of Whisper large-v3 on 234 public YouTube videos in 9 languages, scored against the caption track each channel uploaded. ## Privacy Glasp does not sell users' highlights, and nothing members save on Glasp is shared through this page. The figures above are aggregate shares only. ## Contact If you work on AI research or evaluation and would like to talk with the Glasp team, write to hi@glasp.co. https://glasp.co