Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | Python (CPython) | CUDA Runtime | llama.cpp | Text Generation Inference | KoboldCpp |
|---|---|---|---|---|---|
| Score | B71 | B-65 | C+63 | F39 | F38 |
| Type | Language | ||||
| Execution | interpreted | aot | aot | hybrid | hybrid |
| Interface | cli | sdk | cli | api | gui |
| Cold Start | 50ms | 100ms | 100ms | 10000ms | 1500ms |
| Memory | 15MB | 500MB | 50MB | 2000MB | 400MB |
| Startup | 10ms | 50ms | 10ms | 5000ms | 300ms |
| Isolation | process | hardware | process | container | process |
| Maturity | production | production | production | production | stable |
| Languages | Python | C, C++, Python | C, C++ | Rust, Python | C++, Python |
| License | Other | Proprietary | MIT | Apache-2.0 | AGPL-3.0 |
| Links |