Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | Python (CPython) | llama.cpp | ROCm | ONNX Runtime | Ollama | KoboldCpp |
|---|---|---|---|---|---|---|
| Score | B71 | C+63 | C56 | C-50 | D49 | F38 |
| Type | Language | |||||
| Execution | interpreted | aot | aot | hybrid | hybrid | hybrid |
| Interface | cli | cli | sdk | sdk | cli | gui |
| Cold Start | 50ms | 100ms | 200ms | 500ms | 1000ms | 1500ms |
| Memory | 15MB | 50MB | 600MB | 300MB | 500MB | 400MB |
| Startup | 10ms | 10ms | 100ms | 100ms | 100ms | 300ms |
| Isolation | process | process | hardware | process | process | process |
| Maturity | production | production | stable | production | production | stable |
| Languages | Python | C, C++ | C, C++, Python | Python, C++, C#, Java | Python, JavaScript, Go | C++, Python |
| License | Other | MIT | MIT | MIT | MIT | AGPL-3.0 |
| Links |