Anemll/Anemll

Health v2 · maintenance only

Flags: prerelease_only no_license

Adoption not part of the score

1647 stars · 77 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

ANEMLL is an open-source Python library and toolchain for converting and running large language models on the Apple Neural Engine (ANE) via CoreML. It includes conversion scripts, a chat CLI, and a reference iOS/macOS/visionOS chat app for on-device LLM inference.

Use cases

• convert llama or qwen models to run on apple neural engine
• run llms locally on iphone or mac with coreml
• benchmark ane inference against huggingface models
• reduce coreml model size with weight deduplication
• evaluate converted models with lm-evaluation-harness
• build an ios chat app with on-device llm

When to choose

• you want to deploy small-to-mid LLMs on Apple Silicon or iOS devices using the ANE
• you need a conversion pipeline from HuggingFace models to CoreML with chunking support
• you want low-power on-device inference instead of cloud APIs

When to avoid

• you target NVIDIA/AMD GPUs or non-Apple hardware
• you need the very latest model architectures not yet supported
• you require a permissively licensed dependency and cannot accept a missing license

Facets

library · maturity active

machine-learning llm-inference llm-training compiler machine-learning large-language-models apple-ecosystem developer-tools python cross-platform apple-neural-engine coreml model-conversion on-device-inference tensor-processors llm-porting chat-app macos ios swift

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Anemll/Anemll")