Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | GGUF | Python (CPython) | Firecracker | CUDA Runtime | AWS Lambda | Node.js | ROCm | KoboldCpp |
|---|---|---|---|---|---|---|---|---|
| Score | A-83 | B71 | B70 | B-65 | B-65 | B-65 | C56 | F38 |
| Type | Language | MicroVM | Serverless | Language | ||||
| Execution | aot | interpreted | aot | aot | hybrid | jit | aot | hybrid |
| Interface | embedded | cli | api | sdk | platform | cli | sdk | gui |
| Cold Start | <1ms | 50ms | 125ms | 100ms | 200ms | 50ms | 200ms | 1500ms |
| Memory | 0MB | 15MB | 5MB | 500MB | 128MB | 40MB | 600MB | 400MB |
| Startup | <1ms | 10ms | 100ms | 50ms | 100ms | 20ms | 100ms | 300ms |
| Isolation | process | process | microvm | hardware | microvm | process | hardware | process |
| Maturity | production | production | production | production | production | production | stable | stable |
| Languages | Any | Python | Any | C, C++, Python | JavaScript, TypeScript, Python, Java, Go, Ruby, .NET, Rust | JavaScript, TypeScript | C, C++, Python | C++, Python |
| License | MIT | Other | Apache-2.0 | Proprietary | Proprietary | MIT | MIT | AGPL-3.0 |
| Links |