Cryofm V1
ByteDance's Cryofm V 1 is a flow-based foundation model for 3D cryo-electron microscopy density maps, employing a Hierarchical Diffusion Transformer architecture to learn deep priors of 3D cryo-EM densities. It is genuinely best at supporting various downstream tasks including density map denoising, anisotropy noise correction, and missing wedge inpainting.
Input
Output
Context
-
Max Output
-
Parameters
335.2M
Technical Specifications
Model TypeImage
Context WindowNot available
Max Output TokensNot available
Parameters335.2M
Release DateDec 23, 2025
Training CutoffNot available
Licenseapache-2.0
Open SourceYes
Input Modalities
Image
Output Modalities
Image
Capabilities
Estimated GPU Requirements for Cryofm V1
Estimates based on INT8 quantization. Actual requirements vary by framework and configuration.
Related Tools
Data sourced from official provider APIs and documentation
Last updated: Jun 23, 2026
Start building with the right model.
Automatically route workloads to the right model for every task, every time.