Which AI models do biologists use? AlphaFold, ESM, scGPT and more, by preprints

14.7% of the preprints read so far say they used an AI model for science — AlphaFold most of all; protein structure and design is the biggest area (6.1% of preprints). Updated daily (as of Nov 16)

Counts preprints first posted Apr 2025–Nov 2026; Nov 2026 is still being read.

What do labs use AI models for?

Preprints that say they used a model in each area, and the models used most there.

Protein structure

6.1% of preprints · top three models
AlphaFold5,668
ColabFold838
ProteinMPNN362

Microscopy and imaging

2.9% of preprints · top three models
Cellpose1,301
StarDist424
Segment Anything Model216

Protein sequence models

2.5% of preprints · top three models
ESM-2807
DeepTMHMM258
DeepLoc158

DNA, RNA and variants

1.6% of preprints · top three models
DeepVariant204
SpliceAI185
AlphaMissense180

Clinical and medical

1.1% of preprints · top three models
MedGemma89
SynthSeg87
PubMedBERT81
1.1% of preprints · top three models
scVI451
CellTypist246
scGPT188

Single-cell AI models 

Behaviour and signals

1.1% of preprints · top three models
DeepLabCut631
ICLabel238
SLEAP103

Chemistry and drugs

0.6% of preprints · top three models
ChemBERTa57

The AI models used most

Ranked by the preprints that say they used each, of 112,324 read so far.

#ModelPreprints
1AlphaFoldAI protein structure prediction · github.com/google-deepmind/alphafold5,668
2CellposeCell segmentation model · cellpose.org1,301
3ColabFoldFast AlphaFold structure prediction · github.com/sokrypton/ColabFold838
4ESM-2Protein language model · github.com/facebookresearch/esm807
5DeepLabCutAnimal pose tracking · github.com/DeepLabCut/DeepLabCut631
6scVISingle-cell variational autoencoder · scvi-tools.org451
7StarDistNucleus segmentation model · github.com/stardist/stardist424
8ProteinMPNNProtein sequence design · github.com/dauparas/ProteinMPNN362
9Boltz-2Structure and binding affinity · github.com/jwohlwend/boltz314
10ESMFoldLanguage-model structure prediction · github.com/facebookresearch/esm299
11DeepTMHMMTransmembrane topology prediction · dtu.biolib.com/DeepTMHMM258
12CellTypistAutomated cell type annotation · celltypist.org246
13ICLabelEEG component classifier · github.com/sccn/ICLabel238
14Segment Anything ModelGeneral image segmentation · github.com/facebookresearch/segment-anything216
15DeepVariantVariant calling from sequencing · github.com/google/deepvariant204
16nnU-NetMedical image segmentation · github.com/MIC-DKFZ/nnUNet204
17scGPTSingle-cell foundation model · github.com/bowang-lab/scGPT188
18SpliceAISplicing effect prediction · github.com/Illumina/SpliceAI185
19Chai-1Biomolecular structure prediction · github.com/chaidiscovery/chai-lab181
20AlphaMissenseMissense variant effect prediction · github.com/google-deepmind/alphamissense180
21RFdiffusionDiffusion protein design · github.com/RosettaCommons/RFdiffusion175
22DeepEMhancerCryo-EM map sharpening · github.com/rsanchezgarc/deepEMhancer166
23scANVISemi-supervised cell annotation · scvi-tools.org162
24DeepLocProtein localization prediction · services.healthtech.dtu.dk158
25NetMHCpanMHC peptide binding prediction · services.healthtech.dtu.dk155
26ESM-CProtein sequence models144
27UNIPathology foundation model · github.com/mahmoodlab/UNI135
28ProtT5Protein language model · github.com/agemagician/ProtTrans134
29AlphaGenomeGenome regulatory effect prediction · github.com/google-deepmind/alphagenome132
30GeneformerSingle-cell foundation model · huggingface.co/ctheodoris/Geneformer116
31crYOLOCryo-EM particle picking · cryolo.readthedocs.io/en/stable109
32SLEAPBehaviour and neural signals103
33ESM3Protein sequence models99
34ESM-1bProtein sequence models98
35DINOv2Self-supervised vision model · github.com/facebookresearch/dinov296
36Boltz-1Protein structure and design93
37ProtenixBiomolecular structure prediction · github.com/bytedance/Protenix93
38Evo 2DNA, RNA and variant effects92
39PSIPREDProtein secondary structure prediction · bioinf.cs.ucl.ac.uk/psipred90
40MedGemmaMedical multimodal language model · developers.google.com89

Use of AI models for science has grown since Apr 2025.

The share of each month's preprints that say they used an AI model for science, Apr 2025–Oct 2026, by the month each was first posted.

Which fields use AI models most?

The share of each field's preprints that say they used an AI model for science (fields with 500+ preprints read).

#FieldUse an AI modelPreprints read
1Bioinformatics39.0% 3,1718,125
2Biochemistry33.2% 1,2873,874
3Synthetic Biology28.3% 261921
4Radiology and Imaging23.9% 175733
5Molecular Biology21.2% 7453,521
6Biophysics20.6% 8274,006
7Microbiology18.6% 1,4868,003
8Health Informatics18.4% 3061,661
9Genomics18.3% 6493,548
10Cell Biology18.1% 9115,028
11Pathology17.9% 138772
12Genetic and Genomic Medicine16.7% 2941,756

Rival tools, head to head

Cellpose 1,301
AI models for science
StarDist 424

How does Tnuki know so much?

  1. Preprints analyzed dailyNew bioRxiv and medRxiv preprints are analyzed daily.
  2. Whole paper readWe assess the full text, tables, and supplementary files.
  3. Only real uses countedWe read to make sure the lab actually used the product, not just cited.
How we count

An AI model reads each bioRxiv and medRxiv preprint and lists what its authors used — instruments, kits, services, software, reagents — and how: used in this work, someone else's data, or only mentioned. A preprint counts here once, when the model says the lab used it, if it was first posted in Apr 2025 or later. Names are merged by a second model (versions and spellings into one tool, brands under their owner). We show counts and shares, and the titles of the latest preprints, each linking to the preprint on bioRxiv or medRxiv; we show no other text from a preprint and no author names. Updated daily, most recently on Nov 16, 2026.

Want to reach out to the labs using AI models in biology?

Get the PIs behind the preprints: institutions, labs, and how to reach them.