Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | Metal | ExLlamaV2 | Text Generation Inference |
|---|---|---|---|
| Score | B74 | D47 | F39 |
| Type | |||
| Execution | aot | aot | hybrid |
| Interface | sdk | sdk | api |
| Cold Start | 50ms | 1000ms | 10000ms |
| Memory | 100MB | 300MB | 2000MB |
| Startup | 20ms | 200ms | 5000ms |
| Isolation | hardware | process | container |
| Maturity | production | stable | production |
| Languages | Swift, Objective-C, C++ | Python, C++, CUDA | Rust, Python |
| License | Proprietary | MIT | Apache-2.0 |
| Links |