DeepMind opens AlphaFold 3 code to researchers
Google DeepMind just released the software code for its protein-folding model AlphaFold 3 for non-commercial scientific research. The release follows six months of academic backlash over withholding the code, though accessing the actual model weights still requires an application request.
AlphaFold 3 is Google DeepMind's flagship protein-structure prediction tool. On November 11, DeepMind made the software code available for researchers to download and run for non-commercial scientific applications.
Code is available, but weights still require approval
Six months ago, DeepMind sparked serious backlash across the scientific community after publishing a Nature paper without releasing the underlying code. This update addresses some of those complaints, but it is not a traditional open-source release.
While scientists can now download the software code, it is restricted to non-commercial use. More importantly, the model weights are not included in the public repository; researchers still have to submit an access request to DeepMind to get them.
Why it matters
For academic biologists and non-commercial labs, this makes AlphaFold 3 much more useful by letting researchers inspect the architecture and run non-commercial pipelines locally.
However, biotech startups and commercial pharma remain entirely locked out. In our view, DeepMind found a middle ground to calm academic outrage while protecting its proprietary edge, but keeping the weights behind an application form means the model is still far from truly open.
Sources (1)
- nature.com(opens in a new tab)Originalnature.com