A1NMR uses a Transformer deep learning model to predict solid-state 13C chemical shifts from SMILES strings — faster than DFT, more accurate than DFT.
Traditional NMR prediction requires hours of DFT computation. A1NMR delivers better accuracy in milliseconds.
176 molecules per second on a single RTX 3050. What takes DFT hours, A1NMR does in milliseconds.
MAE=0.476 ppm — more accurate than CSTShift (DFT+GNN hybrid) without any quantum chemistry computation.
Enter a SMILES string, get back predicted chemical shifts and spectrum. No DFT expertise needed. REST API available.
A1NMR achieves the best accuracy-speed tradeoff among all available methods.
| Method | Type | MAE (ppm) | Speed |
|---|---|---|---|
| ★ A1NMR SOTA | Transformer | 0.476 | ~0.005s |
| CSTShift | DFT+GNN | 0.500 | hours |
| CASCADE-2.0 | 3D MPNN | 0.730 | ~1s |
| Gaussian (DFT) | Quantum | 0.4-1.0 | hours-days |
Enter a SMILES string to see A1NMR in action. No signup required.
Sign up to view full spectrum · Free plan includes 50 predictions/month
Start free. Upgrade when you need more.
50 free predictions every month. No credit card required.