SCIME RESEARCH SCORE · COMMUNITY RANKINGS
See how research is taking shape. Share your score. Keep new ideas private and let readers discover your published papers.
2026-10 · Published research · 0 published papers
No scores shared yet. Be the first to share yours.
Evaluate your idea →YOUR IDEA, YOUR CHOICE
Publishing is optional and available to signed-in members, including Free accounts. Anyone can browse. New ideas show your username and score. Published research also shows the paper title and original link. Private notes and research artifacts stay private.
Build your private portfolio →KEEP MAKING PROGRESS
Develop a hypothesis, choose baselines and add evidence. Reassess your research and track how its evaluation changes over time.
New evidence can raise or lower a score.
Experimental AI assessments, not a measure of a researcher’s ability or a guarantee of originality. Only voluntarily published scores from verified accounts are included. Published research ranks each shared paper separately, including multiple papers by the same researcher. Idea stages use each researcher’s highest score. Rankings use the current UTC month and evaluator cohort. Ties share a rank. Published research also requires verified ownership of its connected Google Scholar profile.
Percentiles appear once at least 20 entries participate and include everyone tied at or above that score. They describe this participating group, not all research ideas. All eligible entries can be browsed, 50 per page. Ranks and percentiles are calculated across the entire group, not just this page. Initial ideas and research with added notes are compared separately; notes are user-provided, not independently validated.
SCIME-Research-0.3 · claude-haiku-4-5. Published PDFs use five weighted dimensions: research question 15%, method and study design 25%, comparisons and validation 20%, results and conclusions 25%, reproducibility documentation 15%. External literature search is not required. Idea assessments use six equally weighted dimensions: Hypothesis, Literature validation, Experimental design, Baselines, Results and Reproducibility. Not-started stages contribute zero; insufficient evidence leaves the aggregate unassessed. Identical saved research states reuse their assessment; literature and model changes can still affect later evaluations.
Automatic uses a server-configured model for each task. Select a model to use it for generation, editing, reviews and assessments. Your selection is saved to your account.
What is a credit?
Credits measure AI usage. Credits pay for AI token usage; one credit is not one AI task.
A short text edit uses fewer credits than reviewing a full paper. The amount depends on text length and the AI model.
Reading, collaboration, reference search and PDF compilation do not use AI credits.
Credits are reserved before AI starts and adjusted to reported usage. If an interrupted request has no usage report, its reservation is retained. The selected credit source pays for your next AI request.
Upgrade or top up credits · /pricing