scientific
AlphaFold 3
Biomolecular structure model. At CASP16 the AF3 server was comparable to, not ahead of, the best groups (B8); AF3 was the core engine of most top groups.
Levelmodel
Classscientific
Parametersnot in the notes
Contextnot in the notes
Released8 May 2024
Opennessopen weights restricted
Runs onsingle datacentre GPU
Statusactive
Benchmark Results
vendor-reported
PoseBustersdomain
Beats Vina docking (P = 2.27e-13); success rate shown only in a figurevendor-reported
no score in the notes
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
successor
AlphaFold 2 is the parent (successor)high confidencenotesWhere this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 2 models. Confidence and evidence are the notes' own.
Industry Use Cases
Healthcare
Protein structure and complexesR&D Topics
Identity
- Developer
- Google DeepMind, Isomorphic Labs
- Family
- not in the notes
- Released
- 8 May 2024 sourcesannouncement page
- Weights published
- –
- Record checked
- 10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- diffusion Diffusion over atom coordinates
- Total parameters
- – sources
- Active parameters
- –
- Checkpoint parameters
- –
- Context length
- not in the notes
- Input
- molecule
- Output
- molecule
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- open weights restricted
- Licence
- AlphaFold 3 model parameters terms of use (Google) sources
- Access routes
- open weights download, API, research only
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
- Runs on
- single datacentre GPU
- Footprint
- Code Apache-2.0; weights only if received directly from Google under separate terms
- Runtimes
- not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Links
ModelAlphaFold 3Google DeepMind, scientific
AlphaFold 3Google DeepMind, scientificWho Runs It
PfizerContract signed, 8 Jan 2026Deployments of these use cases; none names this model.
PfizerContract signed, 8 Jan 2026Deployments of these use cases; none names this model.Use cases 1
Sources
- Open source protein models: OpenFold vs Boltz scalingbiotech.substack.com https://scalingbiotech.substack.com/p/open-source-protein-models-openfold
- GitHub - google-deepmind/alphafold3: AlphaFold 3 inference pipeline. github.com https://github.com/google-deepmind/alphafold3
- AlphaFold3 at CASP16 - PMC ncbi.nlm.nih.gov https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12750027/
- Accurate structure prediction of biomolecular interactions with AlphaFold 3 - Nature nature.com https://www.nature.com/articles/s41586-024-07487-w
- AlphaFold3 Goes Open Source biopharmatrend.com https://www.biopharmatrend.com/news/alphafold3-goes-open-source-1017/
- blog.google/technology/ai/google-deepmind-isomorphic-alph... https://blog.google/technology/ai/google-deepmind-isomorphic-alphafold-3-ai-model/
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "AlphaFold 3", https://aurion.semai.id/models/google-deepmind-alphafold-3/, records of October 2026.