Thank you for your interest in our emerging DNA basecalling model, the hyper model, announced at London Calling 2025. The hyper model leverages the compute features of modern GPUs to harness the potential of nanopore signal and deliver the highest single-read accuracy.
This model has significantly increased compute requirements compared to existing HAC and SUP basecalling models and is therefore not intended for routine or real-time workflows. The hyper model is best suited for offline analysis on datasets where single-read accuracy is critical. This model currently does not support calling base modifications.
The hyper model has been optimised to run on advanced NVIDIA hardware, requiring NVIDIA Hopper or Blackwell GPUs with ≥32 GB of VRAM. Multi-GPU configurations are recommended for best performance.
If your work requires the highest single-read accuracy and is supported by suitable GPU infrastructure, we invite you to register your interest using the form below. If you are selected to participate, we will contact you with further information. Please contact us at ea_programs@nanoporetech.com if you need any assistance.
