Qwen2.5-Omni-3B-Audio-CoreAI

Core AI is Apple's on-device ML runtime in iOS 27 / macOS 27 and the successor to Core ML: PyTorch models are exported with Apple's coreai-torch (LLMs: coreai.llm.export) into .aimodel bundles that run on the GPU or the Neural Engine, e.g. Qwen3-8B 4-bit decodes at 94 tok/s on an M4 Max GPU, MLX 90 under the same protocol (apple-silicon-llm-bench, macOS 27 beta 26A5353q, 2026-06-11).

<!-- gen-cards:devicemark begin (managed by scripts/gen-cards + tools/devicemark_row.py — edit cards.json, not this block) -->

This model has no row on DeviceMark, the on-device LLM leaderboard.

<!-- gen-cards:devicemark end -->

Qwen2.5-Omni-3B Audio Understanding — Core AI

Qwen2.5-Omni-3B's Thinker converted to Apple Core AI (.aimodel / .aimodelc, iOS 27 / macOS 27) for on-device audio understanding — the model describes the sounds it hears (events, texture, emotion, music), it is not a transcriber. "I hear a loud hissing sound." · "…a continuous sine wave sound." · "…a series of beeps."

Part of the CoreAI-Model-Zoo. Device-verified on iPhone 17 Pro (A19 Pro) and M4 Max.

<!-- gen-cards:use-it begin id=qwen2.5-omni-3b-audio (managed by scripts/gen-cards — edit cards.json / QuickStart.swift, not this block) -->

Use it

New to Core AI? Start with CoreAIKit 0.7.3. Follow its requirements and first-run steps for qwen3-0.6b, then open the same release's ChatDemo. The README records the tested OS/SDK and download size; model and device coverage is stated per example.

⚡ One line — this model is the default behind the kit's task op (import CoreAIOps; no session, no model plumbing, downloads on first use):

let scene = try await CoreAI.describeAudio(audioURL)

Every op, one shape — Cookbook.

▶️ Run it (source) — the AudioChat runner (GUI + CLI, one app for every audio-understanding model in the catalog):

git clone --branch 0.7.3 --depth 1 https://github.com/john-rocky/coreai-kit
export DEVELOPER_DIR=/Applications/Xcode-27.0.0-RC.app/Contents/Developer
open -a /Applications/Xcode-27.0.0-RC.app coreai-kit/Examples/AudioChat/AudioChat.xcodeproj
# → Run, then pick "Qwen2.5-Omni 3B Audio" in the model picker

# agents / headless (macOS):
cd coreai-kit/Examples/AudioChat
swift run -c release audiochat-cli --model qwen2.5-omni-3b-audio --audio sample.wav --prompt "What do you hear?"

Use Xcode build 27A266a from the release's .xcode-pin; adjust the app path if your installation is named differently.

💻 Build with it — complete; the glue is kit API, copy-paste runs:

import CoreAIKit
import FoundationModels

let audio = try await KitAudioModel(catalog: "qwen2.5-omni-3b-audio")
try await audio.attach(samples: AudioFile.pcm16kMono(audioURL))  // clip → encoder buffer
let session = LanguageModelSession(model: audio)
let reply = try await session.respond(to: question)
// reply.content: what the model heard, described fully on-device

The take-home is Examples/AudioChat/Sources/QuickStart.swift — this exact code as one typed function, no UI; the CLI is an argument shell over it, and the GUI drives the same KitAudioModel(catalog:) behind a LanguageModelSession. Live mic? MicRecorder (kit API) captures 16 kHz mono [Float] — attach that instead. One clip per session; attach a new clip to ask about different audio.

Integration checklist

• SPM: https://github.com/john-rocky/coreai-kit (exact 0.7.3) → product CoreAIKit
• Info.plist: NSMicrophoneUsageDescription — only if you record
• Entitlements: none needed (macOS)
• First run downloads the model — ~5,485 MB (Mac) — then it loads from the local cache (Application Support; progress via the downloadProgress callback)
• Measure in Release — Debug is ~3× slower on per-token host work

<!-- gen-cards:use-it end -->

What's here

Two models, run as a pair on the coreai-pipelined GPU engine:

The decoder's audio embeds ride one static-input buffer (audio_embeds [750,2048]); the prompt's <|AUDIO|> placeholders carry extension ids vocab + slot the graph gathers. TMRoPE collapses to 1-D for audio+text, so positions are engine-native (no rope-shift inputs). iPhone uses the AOT decoder so the 3.9 GB graph dodges the on-device JIT jetsam; the AOT weights mmap as clean pages, so it loads comfortably (≈5.9 GB headroom after load on a 12 GB device, with the increased-memory-limit entitlement).

Use it

The coreai-audio app (record from the mic / choose a file / demo clip → "what do you hear?"), or CoreAIKit:

let model = try await KitAudioModel(model: .qwen2_5Omni3B)   // downloads decoder + encoder
try await model.attach(samples: pcm16kMono)                  // mel → encoder → static buffer
let answer = try await LanguageModelSession(model: model).respond(to: "What do you hear?")

The 16 kHz log-mel front end is Whisper-large-v3 (Accelerate/vDSP), bit-exact with the HF feature extractor (gated cos 1.0). Any clip is decoded to 16 kHz mono, ≤ ~30 s.

Conversion / numerics

Conversion code + gates: conversion (export_qwen2_5_omni_thinker.py / export_qwen2_5_omni_audio.py). Decoder int8lin gates top-1-exact vs the fp32 HF oracle; the encoder static rework is cos 1.0 vs eager (GPU 0.99999); the Swift vDSP mel is cos 1.0 vs the HF extractor. iPhone greedy matches the Mac content (white-noise → "I hear a loud hissing sound.").

License

Apache-2.0 (inherits Qwen2.5-Omni-3B). A community conversion — not affiliated with Alibaba or Apple.

<!-- funnel:v1 -->

---

More models in this format: Core AI Model Zoo — 75 models, each with the recipe that produced it.

Want a different model on-device? Open a request — free, open weights only; the export and its measured numbers get published publicly.

<!-- /funnel:v1 -->