Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | ONNX Runtime | ExLlamaV2 | KoboldCpp |
|---|---|---|---|
| Score | C-50 | D47 | F38 |
| Type | |||
| Execution | hybrid | aot | hybrid |
| Interface | sdk | sdk | gui |
| Cold Start | 500ms | 1000ms | 1500ms |
| Memory | 300MB | 300MB | 400MB |
| Startup | 100ms | 200ms | 300ms |
| Isolation | process | process | process |
| Maturity | production | stable | stable |
| Languages | Python, C++, C#, Java | Python, C++, CUDA | C++, Python |
| License | MIT | MIT | AGPL-3.0 |
| Links |