Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | Python (CPython) | CUDA Runtime | Node.js | llama.cpp | ROCm | KoboldCpp | LM Studio |
|---|---|---|---|---|---|---|---|
| Score | B71 | B-65 | B-65 | C+63 | C56 | F38 | F33 |
| Type | Language | Language | |||||
| Execution | interpreted | aot | jit | aot | aot | hybrid | hybrid |
| Interface | cli | sdk | cli | cli | sdk | gui | gui |
| Cold Start | 50ms | 100ms | 50ms | 100ms | 200ms | 1500ms | 2000ms |
| Memory | 15MB | 500MB | 40MB | 50MB | 600MB | 400MB | 800MB |
| Startup | 10ms | 50ms | 20ms | 10ms | 100ms | 300ms | 500ms |
| Isolation | process | hardware | process | process | hardware | process | process |
| Maturity | production | production | production | production | stable | stable | stable |
| Languages | Python | C, C++, Python | JavaScript, TypeScript | C, C++ | C, C++, Python | C++, Python | Python |
| License | Other | Proprietary | MIT | MIT | MIT | AGPL-3.0 | Proprietary |
| Links |