Review a PubMed paper with AI

Use a PubMed record URL or PMID rather than a search-results page. You can also use a DOI. Confirm the title, authors and year after retrieval, especially when several papers report results from the same trial. MedLit attempts to retrieve bibliographic data and legally available text.

Open-access full text may be available through PubMed Central or Europe PMC. Some papers provide only an abstract. A provider connection does not unlock a publisher subscription or recover missing methods. MedLit retains retrieval warnings so the report reflects the text that was actually reviewed.

Connect an OpenAI, Anthropic Claude or Google Gemini API key directly. OpenRouter remains optional. Load the model list or enter an exact text-model ID, then run the connection test. Your provider bills the usage. Restricted permissions, exhausted credits or unavailable models produce an error instead of a fabricated report.

Review the quotations, uncertain judgments and limitations in the completed appraisal. With saving enabled, the Library keeps the report for later reading without another model call. Export a backup or a citation for your reading workflow. A model-generated appraisal is a starting point for source checking, not a PubMed endorsement.

Explore the teaching example without an account, or bring a paper and connect your own AI provider.