Loading…
Bio platform
NVIDIA BioNeMo / NIM
GPU-accelerated inference for structure prediction, protein design, docking and genomics, as hosted endpoints or self-hosted microservices.
- Models served
- 18
- Offers
- InferenceWeights
- Coverage
- Protein 10Small molecule 5Imaging 3DNA & Gene 1+1 more
- Deployment
- Hosted API · or self-host
- Pricing
- Free dev tier · per-GPU licence
Models you can run on NVIDIA BioNeMo / NIM
Most cited first38.6K—14.8K
Protein structure prediction model that folds amino acid sequences into 3D structures with atomic accuracy, scoring a median GDT of 92.4 at CASP14.
Protein
5.4K2.2M4.2K
Protein structure prediction from a single sequence, with no multiple sequence alignment. Folds a 384-residue protein in 14.2 seconds on one GPU.
Protein
3.3K—14.8K
Protein complex structure prediction model extending AlphaFold 2 with paired MSA processing and ipTM scoring for multi-chain, multimeric assemblies.
Protein
2K—1.8K
Message passing neural network for fixed-backbone protein sequence design. Achieves 52.4% native sequence recovery, far surpassing Rosetta's 32.9%.
Protein
1.5K—3K
De novo protein design diffusion model that generates backbone structures conditioned on binding targets, symmetry constraints, and functional motifs.
Protein
How access works
Two routes exist. Hosted endpoints require an account and an API key and are the fastest way to evaluate — browser use draws no trial credits, only remote API calls do — and run under trial service terms, with each model carrying its own licence. The other route takes the microservice and runs it on-premises or in a private cloud. Research use is covered by developer program membership; production self-hosting requires an enterprise licence. Residency for the hosted endpoints is undocumented.
- Self-serve
- Trial credits
- Public price list
- API access
- Self-hostable
What it costs
The production rate is published. Self-hosted microservices are licensed per GPU rather than per model or per microservice, so cost scales with the hardware pointed at them rather than the number of models run — which favours consolidating workloads onto one fleet. A time-limited evaluation licence is available on request. Before that, hosted endpoints allow genuinely free evaluation, and the underlying framework is free to use in its own right; the licence buys support.
Last verified
