Runtime comparison
ExLlamaV2 vs MLX LM
ExLlamaV2 is a MIT-licensed runtime for Linux · Windows running on CUDA · ROCm; MLX LM is MIT-licensed for macOS on Metal. MLX LM rates easier to set up (4/5 vs 3/5 in our sourced ratings). Pick ExLlamaV2 for maximum control over how the model runs; pick MLX LM for maximum control over how the model runs.
| Spec | ExLlamaV2 | MLX LM |
|---|---|---|
| License | MIT | MIT |
| Operating systems | Linux · Windows | macOS |
| GPU backends | CUDA · ROCm | Metal |
| Install | pip / source | pip |
| Ease of use | 3/5 | 4/5 |
| Quant formats | exl2, gptq | mlx |
| Engine | exllamav2 | mlx |
| GitHub stars | 4,581 | 6,236 |
| Latest version | v0.3.2 | v0.31.3 |
| Graphical app (GUI) | No | No |
| Server mode | No | Yes |
| OpenAI-compatible API | No | Yes |
| CPU offload | No | No |
| Multi-GPU | Yes | No |
| Speculative decoding | Yes | No |
| LoRA support | Yes | Yes |
| KV-cache quantization | Yes | Yes |
Which should you use?
ExLlamaV2
Best for maximum control over how the model runs. Command line and server, runs on CUDA · ROCm.
Full ExLlamaV2 guide →MLX LM
Best for maximum control over how the model runs. Command line and server, runs on Metal.
Full MLX LM guide →Neither runtime changes whether a model fits — that is your memory and quantization. Check your hardware first, then pick the runtime.
Sources: ExLlamaV2 github.com/turboderp-org/exllamav2 (as of 2026-07-08); MLX LM github.com/ml-explore/mlx-lm (as of 2026-07-08).