Latent Labs deploys NVIDIA Blackwell B300 GPUs with Nebius for training and inference in its generative molecular-biology models. These models support protein design and drug-discovery research for industry and academic partners. Its March 2026 announcement describes deployment on Nebius’s full-stack AI infrastructure as demand for compute grew. A Nebius representative describes helping the team test Blackwell after it had used H100 GPUs for model training.
Latent Labs reports initial training speedups of 2–3x, exceeding 4x after hardware-specific optimizations for Blackwell. It identifies NVIDIA Blackwell Server Edition GPUs for inference and reports more than 60% faster inference relative to previous-generation chips, shortening timelines for partners using its models in active drug-discovery programs.
CEO Simon Kohl describes a compute-diverse strategy to meet partners’ needs across industry and academia. The teams are exploring closer integration between Latent Labs’ models, NVIDIA’s hardware roadmap and Nebius AI Cloud for model training and inference.
Latent Labs reports deploying molecular-biology models on Nebius’s full-stack AI infrastructure using NVIDIA Blackwell B300 GPUs. Initial training speedups of 2–3x exceeded 4x after hardware-specific optimizations. Inference was more than 60% faster than on previous-generation chips. The teams are exploring closer integration with NVIDIA’s hardware roadmap and Nebius AI Cloud for further training and inference gains.
Latent Labs announcement16 Mar 2026 · latentlabs.com · Web pageLatent Labs reports deploying molecular-biology models on Nebius’s full-stack AI infrastructure using NVIDIA Blackwell B300 GPUs. Initial training speedups of 2–3x exceeded 4x after hardware-specific optimizations. Inference was more than 60% faster than on previous-generation chips. The teams are exploring closer integration with NVIDIA’s hardware roadmap and Nebius AI Cloud for further training and inference gains.
Why AI Medical Research Needs Vast Computing Power and Where Nebius AI Cloud Fits In - Beyond Longevity
14 Oct 2026 · beyond-longevity.co.uk · Web page
A Nebius representative says Latent Labs used H100 GPUs for model training and tested Blackwell GPUs with Nebius, reporting a fourfold increase in training-cluster performance.
Why AI Medical Research Needs Vast Computing Power and Where Nebius AI Cloud Fits In - Beyond Longevity14 Oct 2026 · beyond-longevity.co.uk · Web pageA Nebius representative says Latent Labs used H100 GPUs for model training and tested Blackwell GPUs with Nebius, reporting a fourfold increase in training-cluster performance.
Latent Labs develops generative AI models and a browser-based platform for protein design, serving academic researchers, biotechnology companies and pharmaceutical teams. Its Latent-X1 and Latent-X2 models generate protein binders, including antibodies, macrocyclic peptides and mini-binder proteins. Latent-Y adds an AI agent that takes written research goals or scientific publications and coordinates target research, candidate generation, computational validation and refinement. The platform offers approved researchers free daily designs and paid additional credits, while enterprise partnerships support commercial drug discovery programs.
Latent Labs deploys NVIDIA Blackwell B300 GPUs with Nebius for training and inference in its generative molecular-biology models. These models support protein design and drug-discovery research for industry and academic partners. Its March 2026 announcement describes deployment on Nebius’s full-stack AI infrastructure as demand for compute grew. A Nebius representative describes helping the team test Blackwell after it had used H100 GPUs for model training.
Latent Labs reports initial training speedups of 2–3x, exceeding 4x after hardware-specific optimizations for Blackwell. It identifies NVIDIA Blackwell Server Edition GPUs for inference and reports more than 60% faster inference relative to previous-generation chips, shortening timelines for partners using its models in active drug-discovery programs.
CEO Simon Kohl describes a compute-diverse strategy to meet partners’ needs across industry and academia. The teams are exploring closer integration between Latent Labs’ models, NVIDIA’s hardware roadmap and Nebius AI Cloud for model training and inference.
Latent Labs reports deploying molecular-biology models on Nebius’s full-stack AI infrastructure using NVIDIA Blackwell B300 GPUs. Initial training speedups of 2–3x exceeded 4x after hardware-specific optimizations. Inference was more than 60% faster than on previous-generation chips. The teams are exploring closer integration with NVIDIA’s hardware roadmap and Nebius AI Cloud for further training and inference gains.
Latent Labs announcement16 Mar 2026 · latentlabs.com · Web pageLatent Labs reports deploying molecular-biology models on Nebius’s full-stack AI infrastructure using NVIDIA Blackwell B300 GPUs. Initial training speedups of 2–3x exceeded 4x after hardware-specific optimizations. Inference was more than 60% faster than on previous-generation chips. The teams are exploring closer integration with NVIDIA’s hardware roadmap and Nebius AI Cloud for further training and inference gains.
Why AI Medical Research Needs Vast Computing Power and Where Nebius AI Cloud Fits In - Beyond Longevity
14 Oct 2026 · beyond-longevity.co.uk · Web page
A Nebius representative says Latent Labs used H100 GPUs for model training and tested Blackwell GPUs with Nebius, reporting a fourfold increase in training-cluster performance.
Why AI Medical Research Needs Vast Computing Power and Where Nebius AI Cloud Fits In - Beyond Longevity14 Oct 2026 · beyond-longevity.co.uk · Web pageA Nebius representative says Latent Labs used H100 GPUs for model training and tested Blackwell GPUs with Nebius, reporting a fourfold increase in training-cluster performance.
Latent Labs develops generative AI models and a browser-based platform for protein design, serving academic researchers, biotechnology companies and pharmaceutical teams. Its Latent-X1 and Latent-X2 models generate protein binders, including antibodies, macrocyclic peptides and mini-binder proteins. Latent-Y adds an AI agent that takes written research goals or scientific publications and coordinates target research, candidate generation, computational validation and refinement. The platform offers approved researchers free daily designs and paid additional credits, while enterprise partnerships support commercial drug discovery programs.
