mlboydaisuke/Qwen3-ASR-1.7B-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).

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

Qwen3-ASR-1.7B — Core AI

Qwen3-ASR-1.7B speech-to-text converted for Apple Core AI, running on-device (iPhone + Mac). The zoo's first ASR model: an AuT audio encoder feeding a Qwen3 decoder on the pipelined engine (audio embeds bound to one static input buffer; {lang}<asr_text>{text} output). ≤30 s clips, 52 languages, automatic language detection.

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 — run the kit's task op on this model (import CoreAIOps; no session, no model plumbing, downloads on first use):

let text = try await CoreAI.transcribe(audioURL, options: .model("qwen3-asr-1.7b"))

Every op, one shape — Cookbook.

▶️ Run it (source) — the Transcribe runner (GUI + CLI, one app for every speech-to-text 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/Transcribe/Transcribe.xcodeproj
# → Run, then pick "Qwen3-ASR 1.7B" in the model picker

# agents / headless (macOS):
cd coreai-kit/Examples/Transcribe
swift run -c release transcribe-cli --model qwen3-asr-1.7b --audio sample.wav

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

let transcriber = try await KitTranscriber(catalog: "qwen3-asr-1.7b")
let samples = try AudioFile.pcm16kMono(url)  // any wav/m4a/mp3 → 16 kHz mono Float
let result = try await transcriber.transcribe(samples: samples)
// result.text, result.language (52 languages)

The take-home is Examples/Transcribe/Sources/QuickStart.swift — this exact code as one typed function, no UI; both the runner's GUI and its CLI call it. Recording? MicRecorder (kit API) captures mic audio as 16 kHz mono [Float] — the record button and permission prompt are your app's own chrome.

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 — ~3,102 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

Driven by CoreAIKit KitASRModel:

let asr = try await KitASRModel(model: .qwen3ASR1_7B)
let r = try await asr.transcribe(samples: pcm16kMono)   // -> (language, text)

Layout: gpu-pipelined/ holds the decoder bundle (*_decode_int8hu_n390_s1, int8) + the paired AuT encoder (*_audio_encoder_fp16_k30, fp16). Same bundles on iOS and macOS.

App: coreai-audio (Transcribe tab — pick Qwen3-ASR or Whisper large-v3-turbo). Card: zoo/qwen3-asr.md.

---

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.