bio.rodeo
ModelsOrganizationsLeaderboardAbout
bio.rodeo

The authoritative source for evaluating biological foundation models. No hype, just honest analysis.

AboutFAQSubmit a modelContact
© 2026 Pulsatance. All rights reserved. ~
Built by Pulsatance

Leaderboard

The state of biological AI at a glance — which models drive the most academic impact and developer adoption, and how open the field actually is. Across 710 models, with 710 evaluated against the Model Openness Framework.

710
Models tracked
125.7K
Total citations
17%
Reach Open Model (III+)
121 of 710 models
3
Fully Open Science (I)
all 17 components open

Impact & scale

The biggest and most-cited models — colored by openness so you can see whether scale and openness go together.

Size vs. Impact

How model scale relates to academic citations (log-log). Showing 135 models with both metrics.

ClosedOpenopenness 0–100
Unevaluated

Most Cited

Academic impact by citation count

#ModelOrganizationCitations
1
AlphaFold 2
Google DeepMind36.3K
2
AlphaFold 3
Google DeepMind11.2K
3
ESM-2 & ESMFold
Meta AI4.7K
4
Cellpose
HHMI Janelia Research Campus3.4K
5
AlphaFold-Multimer
Google DeepMind3.1K
6
Big Bird
Google Research2.8K
7
scVI (CELLxGENE Census)
Chan Zuckerberg Initiative2.3K
8
ProteinMPNN
Institute for Protein Design1.8K
9
LLaVA-Med
Microsoft Research1.8K
10
AlphaMissense
Google DeepMind1.6K

Most Starred

Developer adoption by GitHub stars

#ModelOrganizationStars
1
Enformer
Google DeepMind15K
2
AlphaFold-Multimer
Google DeepMind14.7K
3
AlphaFold 2
Google DeepMind14.7K
4
AlphaFold 3
Google DeepMind8.2K
5
BioGPT
Microsoft Research Asia / Microsoft Research4.5K
6
MedSAM
Bowang Lab / University Health Network / University of Toronto / Vector Institute / Western University / New York University / Yale University4.3K
7
ESM-1v
Meta AI4.1K
8
ESM-2 & ESMFold
Meta AI4.1K
9
ESM-1b
Meta AI4.1K
10
Boltz-2
MIT CSAIL / Recursion Pharmaceuticals4K

How open is biological AI?

MOF's pass/fail tiers leave most models unclassified, so the bio.rodeo openness score rates them on a 0–100 gradient — split into usability (can I run it?) and reproducibility (can I retrain it?). These views show that gradient, the trend over time, where models fall short, the licenses in play, and how openness relates to impact.

Usability vs. Reproducibility

Each dot is a model; color is its bio.rodeo openness band. Below the diagonal: easier to run than to retrain (open weights, closed recipe); above it: more reproducible than usable. Showing 710 evaluated models.

reproducible · less usablefully openclosedopen weights · closed recipe
ClosedOpenopenness 0–100
MOF Class I/II/III

Most Open

Top models by bio.rodeo openness score (usability + reproducibility, license-aware).

#ModelUseReproOverall
1
SegVol
Beijing Academy of Artificial Intelligence
100
100
100
2
WaveOrder
Chan Zuckerberg Initiative
99
100
100
3
Cytoland
Chan Zuckerberg Biohub / Mehta Lab
100
95
98
4
Boltz-1
MIT
100
95
97
5
Geneformer
Broad Institute / Dana-Farber Cancer Institute
100
93
96
6
TopCUP
Chan Zuckerberg Initiative
100
92
96
7
scVI (CELLxGENE Census)
Chan Zuckerberg Initiative
100
92
96
8
Aiki-XP
Aikium
99
92
96
9
SAM-Med3D
Shanghai AI Laboratory
100
95
95
10
MonjuDetectHM
Chan Zuckerberg Initiative
100
92
95

Openness Over Time

Evaluated models by release year and overall openness tier (open / partial / closed)

Open
Partial
Closed

Component Openness

Share of models that release each MOF component openly, grouped by pillar — where the field falls short

Documentation
Model Artifacts
Data & Evaluation
Open
Described only
Restricted
Unavailable
Unknown

Openness and Impact

Median citations for models that release a component openly vs. not, sorted by the size of the gap

Component is open
Not open (described / restricted / unavailable)

License Landscape

Most common licenses across released components, colored by whether they are open

Open license
Restrictive license

Who & what

Openness and impact broken down by biological domain and by organization, plus the field's growth in scale over time.

Category Breakdown

Models per biological domain, split into open (Class III+) and not open

Open (Class III+)
Not open

Top Organizations

By total citation count

Google DeepMind
53.5K citations6 models55.1K stars
Meta AI
6.6K citations5 models15.3K stars
HHMI Janelia Research Campus
4.9K citations4 models8.9K stars
Microsoft Research
4K citations12 models7.1K stars
Institute for Protein Design
3.4K citations6 models6.7K stars
Google Research
2.9K citations2 models634 stars
Chan Zuckerberg Initiative
2.4K citations17 models2.2K stars
Stanford University
2.2K citations18 models2K stars
Mahmood Lab
1.4K citations1 model741 stars
Microsoft Research Asia / Microsoft Research
1.4K citations1 model4.5K stars
Bowang Lab / University Health Network / University of Toronto / Vector Institute / Western University / New York University / Yale University
1.3K citations1 model4.3K stars
Rostlab
1.3K citations2 models1.6K stars
UC Berkeley
1.2K citations5 models1.2K stars
Bowang Lab
1.1K citations3 models2K stars
Church Lab
1.1K citations1 model365 stars

Field Timeline

How model scale has grown over time, colored by bio.rodeo openness score. Bubble size reflects citation count where available. Showing 178 models.

ClosedOpenopenness 0–100
Unevaluated