AX-Qwen3-ASR-1.7B-MLX-AXQ-4bit

An AXQuant (AXQ) mixed-precision MLX checkpoint for Apple Silicon, converted directly from the BF16 source model. The language path is quantized while the audio tower are preserved at BF16 in the checkpoint (or a bound sidecar when present).

Development evidence — not a certified AXQuant release. This package has conversion and artifact-integrity records, but it does not publish measured quality, long-context, kernel-speed, or MTP-speed evidence. Do not interpret the AXQ product label as a benchmark claim.

Stable-name v2. main serves the audited v2 artifact for backward compatibility. The same revision is tagged v2; when this repository replaced an earlier artifact, that prior revision remains recoverable at legacy-pre-v2.

Model details

Property	Value
Base model	Qwen/Qwen3-ASR-1.7B
Source revision	7278e1e70fe206f11671096ffdd38061171dd6e5
Product family	qwen3-asr
Source architecture	Qwen3ASRForConditionalGeneration (dense); text path optimized
Main-model parameters	2.04B logical parameters
Quantizer	AXQuant 1.2.0
Hub budget class	4bit
Artifact edition	v2
AXQuant base precision class	6p9bpw
Planned storage-adjusted BPW	6.9095
Measured main-model BPW	6.9100
Measured total BPW	6.9100
Safetensors weight size	1.76 GB
Approximate complete download	1.77 GB
Configured maximum context	unrecorded tokens; practical limits depend on unified memory
Primary MLX runtime	MLX-Audio
AX Engine native execution	Not established; no validated native manifest is included
MTP present	False
Vision present	False
Audio present	True

This repository contains MLX Safetensors. It does not contain PyTorch or GGUF weights.

Choosing an AXQ pack

AXQ names describe a storage-budget product class, not one uniform precision applied to every tensor. Protected tensors remain at higher precision, so the exact measured BPW is authoritative. In particular, a 6bit-named mixed plan may retain 4bit as its base precision while selecting 6-bit, 8-bit, or BF16 for other tensors to meet an approximately 6-BPW total budget. Protection floors can also raise a 4bit-named pack close to (or above) a 6bit budget on small or heavily protected models.

Sibling	Intended trade-off
4bit sibling	Lower-storage AXQ budget; check its exact BPW
6bit sibling	Higher average precision near the 6-BPW budget

See the AutomatosX MLX model catalog for related MLX and OptiQ alternatives.

Download

python -m pip install -U huggingface_hub
hf download AutomatosX/AX-Qwen3-ASR-1.7B-MLX-AXQ-4bit --local-dir ./AX-Qwen3-ASR-1.7B-MLX-AXQ-4bit

Allow at least 1.77 GB of free disk space. Pin the resulting Hub commit in reproducible deployments rather than relying indefinitely on main.

Run with MLX-Audio

python -m pip install -U mlx-audio
python -m mlx_audio.stt.generate \
  --model AutomatosX/AX-Qwen3-ASR-1.7B-MLX-AXQ-4bit \
  --audio ./audio.wav \
  --output-path ./transcript \
  --format txt

The protected audio tower and AXQ language decoder are loaded together by MLX-Audio. The artifact records MLX 0.32.0; conversion used the Qwen3-ASR implementation supplied by MLX-Audio.

AX Engine status

This package does not include a validated native model-manifest.json, so AX Engine execution is not established by this release. The AX Engine fields in axquant_runtime.json describe the intended compatibility contract, not observed runtime evidence. Use the architecture-specific MLX runtime path above. The artifact records AX Engine version not recorded, but version discovery alone is not a runtime check.

Quantization layout

Main-weight precision	Parameters	Share
4bit	1.41B	69.15%
8bit	311.16M	15.27%
bf16	317.60M	15.58%

• Quantization methods: affine, bf16.
• Group sizes used by quantized assignments: 32, 64.
• MTP sidecar: not included.
• Vision sidecar: not included.
• Audio weights: protected BF16 in main shards.
• Optimization scope: text-path.
• Support tier: convertible.

BF16 sidecars, when present, are included in total download size. Their presence does not by itself establish MTP acceleration or vision-language quality.

Evidence and validation status

Check	Status
Planning evidence	architecture_prior
Calibration	none; the allocation is based on architecture priors
Quantizer execution	197/197 recorded module conversions succeeded; 0 fallbacks
AX Engine native manifest	not included
Quality versus BF16 or uniform baselines	Not published; no quality-retention claim
MTP acceptance and speed	not measured; no MTP speedup claim
AX Engine kernel evidence	unmeasured
Vision-language quality	Not applicable (no vision tower in this package)
Speech-recognition quality	Not evaluated or claimed; audio tensors are preserved at BF16
Long-context quality	unrecorded-token capacity is config metadata, not a validated claim
Release certification	Not certified; formal AXQuant M0-M8 gates are not closed

Intended use and limitations

• Intended for local development and evaluation on Apple Silicon with MLX-compatible runtimes.

• No minimum unified-memory figure is claimed; loadability depends on model size, context length, KV-cache policy, runtime buffers, and other processes using unified memory.

• Architecture-prior allocation is not measured sensitivity. It must not be presented as measured model quality.

• Audio weights are preserved at BF16, but this release does not claim measured transcription quality versus the BF16 source.

• The configured context window can require substantially more memory as the KV cache grows.

• AX Engine execution is not established because this package has no validated native manifest.

• Upstream capabilities, limitations, biases, and responsible-use guidance still apply.

Provenance and audit files

• axquant_manifest.json: package identity, byte accounting, runtime contract, software versions, and file checksums.
• axquant_plan.json: per-tensor precision decisions and planning evidence.
• axquant_quantizer_execution.json: conversion coverage and fallback records.
• axquant_runtime.json: declared AX Engine and MLX compatibility metadata; runtime checks remain separate evidence.

All published provenance uses repository-relative paths. Local source paths are stripped before publication. The checkpoint was converted from BF16 rather than re-quantized from an OptiQ artifact. If an OptiQ repository is published separately, it uses a different quantizer and should not be assumed to have identical BPW or quality.

License

The checkpoint follows the upstream model license where applicable (often Apache License 2.0). See the Qwen/Qwen3-ASR-1.7B model card for license terms, model limitations, and responsible-use guidance.