Engine and serving snapshot
| Dimension | macMLX | oMLX |
|---|---|---|
| Platform | Apple Silicon macOS 14 or later | Apple Silicon macOS |
| Core runtime | Swift in-process inference through Apple MLX; the default path requires no Python runtime | Python/FastAPI inference core with native SwiftUI shell |
| Model workflow | Supported MLX language, vision, embedding, LoRA, and checkpoint-governed native model workflows | Multi-model MLX serving |
| Interfaces | SwiftUI app, macmlx CLI, compatible HTTP APIs with structured output, and integrated tool-routing surfaces | Menu-bar app, server tools, OpenAI and Anthropic compatibility |
| Factual focus / audience | Swift-native serving with eligibility-gated continuous batching, LCP prompt reuse, structured output, and speculative decoding | Continuous batching and paged hot/SSD caching with prefix sharing and CoW |
Documented limitations
oMLX
- The stable v0.4.4 snapshot is Apple-Silicon-specific and uses a Python server core rather than in-process Swift inference.
Snapshot
Official sources
- Eligibility-gated continuous batching
- Apple Silicon macOS installation
- Speculative decoding
- Track G tested models
- Eligibility-gated continuous batching
- Local embeddings
- InternLM3 theoretical support
- InternLM3 theoretical support
- OpenAI endpoint compatibility
- Structured output
- Integrated chat tool routing
- Eligibility-gated continuous batching
- Eligibility-gated continuous batching
- Trie longest-prefix reuse
- Speculative decoding
- Paged KV, block sharing, and CoW
- oMLX · github.com
- oMLX · github.com