Axiomi.si
Submit a tool

// category · 6 tools · 0 open source

AI Research Platforms

Where research is published, discussed, reproduced and shared: preprint servers, review platforms, model hubs and hosted notebooks.

Where AI research lives

Machine learning moves faster than traditional journals, so most of the field's work appears first on open platforms. This category collects the places where papers are posted, reviewed, discussed and reproduced, along with the hubs where models and code are shared.

Preprints and peer review

arXiv is the default preprint server for machine learning; new papers usually appear there before or alongside a conference. Reading arXiv alone can be hard work, so discussion layers have grown on top of it. alphaXiv lets readers comment line by line on arXiv papers and lets authors reply. OpenReview runs peer review for conferences such as ICLR and NeurIPS and publishes reviews and author responses next to the paper, which is often the fastest way to see where a paper is weak.

Model hubs and reproducibility

Hugging Face Hub hosts most open weight models and datasets, and its huggingface_hub Python client (version 2.2.0 on PyPI as of October 21, 2026) handles downloads, uploads and authentication. When a paper releases weights, the model card on the Hub usually states the license, the training data and the intended use, which is the information you need before you build on it. Our open weight vs open source explainer shows how to read those licenses.

How to keep up without drowning

A practical routine is to follow a few topics rather than everything. Use arXiv listings or alphaXiv for new work, check OpenReview for the reviews on the papers that matter to you, and look for code and weights on GitHub and Hugging Face before you trust a result. Benchmarks reported in a paper are the authors' own numbers until someone else reproduces them.

For a field specific reading list, our mechanistic interpretability explainer points to the key papers and the Transformer Circuits archive, and the best ML frameworks for Python guide covers the libraries most research code is written in. Each platform page below links to its official site.

Hugging Face Hubhuggingface.coCentral hub for open models, datasets and demo Spaces, the GitHub of the machine learning community.Research PlatformsFreemium★ 4kalphaXivalphaxiv.orgDiscussion layer on top of arXiv where readers comment line by line on papers and authors respond.Research PlatformsFreeHN 7Google Colabcolab.research.google.comHosted Jupyter notebooks with free and paid GPU and TPU access, ideal for quick experiments and teaching.Research PlatformsFreemiumKagglekaggle.comData science platform with public datasets, free notebook compute, model hosting and competitions.Research PlatformsFreeOpenReviewopenreview.netOpen peer review platform used by ICLR, NeurIPS and other venues, publishing reviews alongside papers.Research PlatformsFreearXivarxiv.orgOpen access preprint server where nearly every important machine learning paper appears first.Research PlatformsFree

// guides

Guides to AI research platforms

All guides →

More categories

Open Weight ModelsML FrameworksTraining & Fine tuningInterpretabilityDatasets