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
Protein

DCFold

Tsinghua University

A single-step generative model for protein structure prediction and binder design that reaches AlphaFold3-level accuracy with a claimed ~15x inference speedup.

Released: May 2026

DCFold is a pretrained single-step generative model for protein structure prediction and protein binder design. It is built on the AlphaFold3 architecture but replaces AlphaFold3's iterative diffusion sampling—which requires many sequential denoising steps at inference—with a single forward pass. The authors report that this design reaches AlphaFold3-level accuracy while delivering a claimed ~15x speedup in inference, addressing one of the main practical bottlenecks of diffusion-based structure predictors: the cost of generating high-quality structures one denoising step at a time.

The model's central methodological contribution is a Dual Consistency training framework paired with a Temporal Geodesic Matching (TGM) scheduler. Consistency-style training teaches a generative model to map directly from noise to a clean sample, collapsing an iterative trajectory into one step; DCFold applies this idea to the AlphaFold3 structure module so that a single evaluation produces a usable structure rather than a starting point for further refinement. A notable property is that one checkpoint serves both structure prediction and binder design without task-specific retraining.

DCFold was introduced in a May 2026 arXiv preprint (arXiv:2605.17899) and accepted as an Oral at ICLR 2026. The arXiv listing does not state author affiliations; the work is attributed here to Tsinghua University's Institute for AI Industry Research (AIR) based on senior authors Hao Zhou and Wei-Ying Ma, who lead AIR's protein generative-modeling group (e.g., ProfileBFN, ESM-AA), with co-authors including members of that group. This affiliation is inferred from the authors' group, not stated on the arXiv record.

#Key Features

  • Single forward-pass generation: Produces a full structure in one model evaluation rather than the multi-step iterative denoising used by AlphaFold3, which is the source of the reported inference acceleration.
  • AlphaFold3-level accuracy: The authors report parity with AlphaFold3 on structure-prediction and binder-design benchmarks despite the reduced step count.
  • Claimed ~15x inference speedup: Removing the iterative sampling loop is reported to cut inference time roughly fifteen-fold relative to the iterative baseline.
  • Unified checkpoint: A single pretrained model handles both structure prediction and binder design without re-training a separate model per task.
  • Dual Consistency + TGM: The training framework combines a dual consistency objective with a Temporal Geodesic Matching scheduler to distill the iterative generative process into one step.

#Technical Details

DCFold inherits the AlphaFold3 architecture, including its representation trunk and diffusion-style structure generation module, and changes how that generative module is trained and queried. Rather than learning a step-by-step denoiser sampled iteratively at inference, the Dual Consistency framework trains the model to be self-consistent across the generative trajectory so that a single step yields a high-quality structure, while the Temporal Geodesic Matching scheduler governs how points along that trajectory are matched during training. The headline empirical claims are AlphaFold3-comparable accuracy on structure-prediction and binder-design evaluations together with an approximately 15x reduction in inference cost from the single-pass design. The arXiv abstract does not describe the training datasets, parameter count, or detailed benchmark tables, so those specifics are not characterized here.

#Applications

DCFold targets workflows where the throughput of AlphaFold3-class predictions matters: large-scale proteome or variant screening, high-throughput binder and interface design campaigns, and any pipeline where iterative diffusion sampling is a latency or compute bottleneck. Because one checkpoint covers both folding and binder design, it can serve as a single backend for groups that would otherwise maintain separate predictive and generative-design models, benefiting protein engineers, structural biologists, and computational drug-discovery teams.

#Impact

DCFold contributes to a broader effort to make AlphaFold3-quality structure prediction cheaper and faster by distilling iterative diffusion into single-step generation, an approach that, if its accuracy claims hold up under independent evaluation, could lower the cost of running modern structure predictors at scale. Its acceptance as an ICLR 2026 Oral signals strong peer interest in consistency-based acceleration of biomolecular generative models. As of this writing, however, no code, model weights, or HuggingFace release has been linked, and the training data is not described in the preprint, so independent reproduction and downstream adoption remain to be established.

Citation

Preprint

DOI: 10.48550/arXiv.2605.17899

DOI: 10.48550/arXiv.2605.17899

Openness

Unclassified
Missing required components

Tags

binder_designdiffusionflow_matchinggenerativestructure_prediction

Resources

Research Paper