videlalvaro/ane-book

Score7.8

Popularity32.0

Risknone

TierSilver

Score breakdown

Usefulness8.0

Novelty9.0

Momentum7.0

Maturity6.3

Open-source/build8.4

Evidence8.0

Workflow potential8.9

Setup ease6.4

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

Useful for AI engineers and Apple-platform developers who want production LLM inference running entirely on the Apple Neural Engine — ane-book is the practitioner's guide with converters, Swift runtimes, and validated model manifests, and every model in the repo runs 100% on the ANE (verified with MLComputePlan, no GPU fallback, no CPU matmuls); for engineering teams that need to deploy LLMs on Ap

Who should use it

AI engineers and Apple-platform developers who want production LLM inference running entirely on the Apple Neural Engine — ane-book is the practitioner's guide with converters, Swift runtimes, and validated model manifests, and every model in the repo runs 100% on the ANE (verified with MLComputePlan, no GPU fallback, no CPU matmuls)Engineering teams that need to deploy LLMs on Apple Silicon hardware (M-series Mac, M4 Max with 48 GB unified memory) without depending on a discrete GPU or a cloud endpoint — the ANE is the dedicated ML accelerator on every Apple Silicon chip and ANE inference does not require external servicesEngineering teams evaluating model selection for ANE — Phi-4-mini-instruct 3.8B at ~17 tok/s (dense general-purpose), Hy-MT 1.5 1.8B at ~34 tok/s (translation), ZAYA1-8B MoE at ~9 tok/s (open MoE), Privacy Filter MoE NER at ~24.6 sent/s (PII detection) cover a useful spreadSwift / Xcode developers who want to integrate ANE inference into a native iOS / macOS app — the Swift runtime is the production-grade surface, no Python wrapper around a one-shot demoEngineering teams that need a verifiable proof of 100% ANE execution — `MLComputePlan` is the verification surface documented in the repoEngineering teams evaluating the ANE as a serious deployment target — the four validated models and measured tok/s are the real benchmark surface (vs vague 'ANE is faster than CPU' claims from vendor blogs)Evaluation: clone the repo, follow the converter path on a model of interest (Phi-4-mini is the recommended starting point), confirm with `MLComputePlan` that execution is 100% ANE, run the Swift runtime on the documented hardware (M4 Max 48 GB, macOS 15, Xcode 16), compare the measured tok/s to the published numbers

Who should skip it

Skip videlalvaro/ane-book if the source link, documentation, or setup requirements do not align with your current workflow or stack.

About this signal

videlalvaro/ane-book is tracked by RepoRadar as a production llm inference on appl in the MIT practitioner's guide to production LLM infer section. It was first seen on 2026-06-25 and last updated on 2026-06-25. The current verdict is 'try now' with a Silver tier and moderate setup difficulty. videlalvaro/ane-book leads on novelty (9.0) and workflow potential (8.9); its lowest signal is maturity (6.3), so factor that in before investing setup time. This page summarizes the evidence RepoRadar has captured from captured source metadata. The score, tier, risk label, and verdict on this page are never influenced by sponsorship, ads, or tips — they reflect only the usefulness, popularity, novelty, momentum, maturity, and evidence signals described in the RepoRadar methodology.

How this item is evaluated

RepoRadar assigned videlalvaro/ane-book a composite score of 7.8 out of 10, placing it in the Silver tier. This score combines weighted sub-signals: usefulness (35%), novelty (18%), momentum (14%), maturity (10%), open-source/build quality (7%), evidence quality (6%), workflow potential (6%), and setup ease (4%). Popularity is tracked separately at 32.0 and never affects the composite score or tier. The risk label of 'none' reflects inherent user-impacting hazards, not generic novelty. Items with no risk flag may still require normal code review before production use.

Putting this into practice? Read How to evaluate an AI tool before you adopt it for the checklist behind this score.

Risk explanation

No inherent user-impacting risk is flagged from the captured evidence.

Evidence links

github.com

Closest alternatives / related signals

ane-bookvidelalvaroalvaro-videlaapple-neural-engineaneproduction-llm-inferenceane-inferenceconverters