Skip to content
AI Atlas
ModelActiveOpen weights

Qwen2-Audio

Qwen Teamfamily · Qwen2qwen.ai

Open in Graph
data quality59

Updated 3 h ago · first seen 12 Sept 2026

model_01M29XAH078M6SSDF8DVQHM9AD

Overview

Identity

Canonical model
Yesidentity confidence: highOne row per real model release. Artifacts (checkpoints, quantisations, conversions) and folded evaluation variants point here.
Official checkpoints
None recordedofficial_checkpoints = hf_repo identifiers carried by the model itself; artifacts are separate entities pointing here through canonical_id.
Artifacts
None recordedSeparate entities (checkpoint · quantization · conversion · packaging) pointing to this model through canonical_id.
Provider deployments
None recorded
API aliases
NoneIdentifiers under which providers and evaluators refer to this model.
Folded evaluation variants
0Effort / thinking variants (…-high, …-non-reasoning) are result configurations of this model, not separate models. Their old URLs redirect here.

Openness

Open weights— weights downloadable; 7 dimensions unknown.

Weights downloadable under a permissive or Creative Commons licence allowing commercial use; code or data may be missing.

  • Weights

    Yes

  • Inference code

    —

  • Training code

    —

  • Training data

    —

  • Dataset

    —

  • Commercial use

    —

  • Redistribution

    —

  • Derivatives

    —

dimensions marked null are unknown, not false

Key facts

Release date

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Status

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Version

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Official page
qwen.ai

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Architecture

Architecture

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Parameters

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Mixture of experts

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Capabilities

Modalities

Modalities
audiotext
Input
audiotext
Output
text

Capabilities

  • Tool calling

    Unavailable

  • Structured output

    Unavailable

  • Reasoning

    Unavailable

  • Vision

    No

    Qwen — official blog · T2

  • Audio

    Yes

    Qwen — official blog · T2

  • Fine-tuning available

    Yes

    Qwen — official blog · T2

Max output

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Languages

Source:Qwen — official blogT2observed 13 h agomediumLLM-extracted

Hardware fit37

Estimated

37 of 37 device × quantization combinations fit.

Run locally: your machine →
Estimated hardware fit
HardwareQuantizationDevice memoryEst. memoryFits
Apple M3 Ultra4bit—4.5 GB est.Yes
Mac Studio (Apple M5 Ultra)4bit—4.5 GB est.Yes
Apple M2 Ultra4bit—4.5 GB est.Yes
Apple M1 Ultra4bit—4.5 GB est.Yes
Apple M3 Max4bit—4.5 GB est.Yes
Apple M4 Max4bit—4.5 GB est.Yes
Mac Studio (Apple M5 Max)4bit—4.5 GB est.Yes
MacBook Pro (Apple M5 Max)4bit—4.5 GB est.Yes
Apple M2 Max4bit—4.5 GB est.Yes
Apple M1 Max4bit—4.5 GB est.Yes
Apple M4 Pro4bit—4.5 GB est.Yes
Mac mini (Apple M5 Pro)4bit—4.5 GB est.Yes
MacBook Pro (Apple M5 Pro)4bit—4.5 GB est.Yes
Apple M3 Pro4bit—4.5 GB est.Yes
Apple M1 Pro4bit—4.5 GB est.Yes
Apple M2 Pro4bit—4.5 GB est.Yes
Apple M44bit—4.5 GB est.Yes
iMac (Apple M4)4bit—4.5 GB est.Yes
Mac mini (Apple M6)4bit—4.5 GB est.Yes
MacBook Air (Apple M5)4bit—4.5 GB est.Yes
MacBook Pro (Apple M5)4bit—4.5 GB est.Yes
Apple M24bit—4.5 GB est.Yes
Apple M34bit—4.5 GB est.Yes
Apple M14bit—4.5 GB est.Yes
NVIDIA GeForce RTX 30904bit24 GB4.5 GB est.Yes
NVIDIA GeForce RTX 40904bit24 GB4.5 GB est.Yes
NVIDIA GeForce RTX 50904bit32 GB4.5 GB est.Yes
NVIDIA A100 80GB4bit80 GB4.5 GB est.Yes
NVIDIA H100 SXM4bit80 GB4.5 GB est.Yes
NVIDIA H100 NVL4bit94 GB4.5 GB est.Yes
NVIDIA DGX Spark4bit128 GB4.5 GB est.Yes
NVIDIA H2004bit141 GB4.5 GB est.Yes
NVIDIA H200 NVL4bit141 GB4.5 GB est.Yes
NVIDIA B2004bit180 GB4.5 GB est.Yes
AMD Instinct MI300X4bit192 GB4.5 GB est.Yes
AMD Instinct MI325X4bit256 GB4.5 GB est.Yes
NVIDIA DGX B2004bit1,440 GB4.5 GB est.Yes
Assumptions (7)
  • Estimated, not measured: weights = parameters × bytes/param × 1.15 runtime overhead (or the observed artifact file size when one is recorded).
  • bytes/param: 4bit = 0.5, 8bit = 1.0, fp16 = 2.0 (uniform quantization, no per-layer exceptions).
  • KV cache: 2 × layers × kv_heads × head_dim × 2 bytes × context × batch when the architecture is known; otherwise 0.5 GB per 8 192 tokens (× batch), independent of architecture (GQA/MLA models need less).
  • A model 'fits' when the estimate is at most the device memory minus 2 GB reserved for the OS and framework.
  • Mixture-of-experts models are estimated on total parameters (all experts must be resident); active parameters are ignored.
  • Device memory uses the largest configuration when several are listed (e.g. Apple silicon tiers).
  • Multi-GPU: device memories are summed; interconnect bandwidth, tensor-parallel replication and pipeline bubbles are not modelled.

Lineage

Open in Graph →
Explicit derived_from / fine_tuned_from / distilled_from relations stated by sources; artifacts collapsed by kind.
ANCESTORS 1QwenQwenQwen2-Audio7B params · this modelQwen2-Audio — 7B params · this model

Versions & Artifacts0

Version history

Max outputfirst observation only

11 Sept 2026current

Opennessfirst observation only

11 Sept 2026current

Parametersfirst observation only

11 Sept 2026current

Statusfirst observation only

11 Sept 2026current

Each hop is a claim: click a value for its source, tier and observation time. Nothing is overwritten — a new observation closes the previous claim.

Artifacts 0

No artifact (checkpoint, quantisation, conversion or packaging) points to this model yet.

Change history19

Temporal, append-only claims: a new observation closes the previous claim instead of overwriting it. Rewind the record with the as-of picker.

Viewing AI Atlas as of 1 Jan 2025 — attributes exactly as the atlas knew them on that day; later corrections are not shown.

Back to today →

Qwen2-Audio was not yet in AI Atlas on 1 Jan 2025

First seen 12 Sept 2026. Nothing is inferred backwards: no attribute is shown for dates before the first observation.
19 claims · 19 properties

Familyfamily1

Claim history for Family
ValueValid from → toStatusSourceConfidenceExtractor
Qwen → currentcurrentQwen — official blogT2mediumllm

Versionversion1

Claim history for Version
ValueValid from → toStatusSourceConfidenceExtractor
2 → currentcurrentQwen — official blogT2mediumllm

Release daterelease_date1

Claim history for Release date
ValueValid from → toStatusSourceConfidenceExtractor
9 Aug 2024 → currentcurrentQwen — official blogT2mediumllm

Statusstatus1

Claim history for Status
ValueValid from → toStatusSourceConfidenceExtractor
active → currentcurrentQwen — official blogT2mediumllm

Opennessopenness1

Claim history for Openness
ValueValid from → toStatusSourceConfidenceExtractor
open-weights → currentcurrentQwen — official blogT2mediumllm

Architecturearchitecture1

Claim history for Architecture
ValueValid from → toStatusSourceConfidenceExtractor
Transformer decoder based on Qwen language model and audio encoder → currentcurrentQwen — official blogT2mediumllm

Parametersparameter_count1

Claim history for Parameters
ValueValid from → toStatusSourceConfidenceExtractor
7B → currentcurrentQwen — official blogT2mediumllm

Mixture of expertsis_moe1

Claim history for Mixture of experts
ValueValid from → toStatusSourceConfidenceExtractor
No → currentcurrentQwen — official blogT2mediumllm

Max outputmax_output_tokens1

Claim history for Max output
ValueValid from → toStatusSourceConfidenceExtractor
256 tokens → currentcurrentQwen — official blogT2mediumllm

Modalitiesmodalities1

Claim history for Modalities
ValueValid from → toStatusSourceConfidenceExtractor
audio, text → currentcurrentQwen — official blogT2mediumllm

Input modalitiesmodalities_input1

Claim history for Input modalities
ValueValid from → toStatusSourceConfidenceExtractor
audio, text → currentcurrentQwen — official blogT2mediumllm

Output modalitiesmodalities_output1

Claim history for Output modalities
ValueValid from → toStatusSourceConfidenceExtractor
text → currentcurrentQwen — official blogT2mediumllm

Languageslanguages1

Claim history for Languages
ValueValid from → toStatusSourceConfidenceExtractor
Cantonese, Chinese, English, French, German, Italian, Japanese, Spanish → currentcurrentQwen — official blogT2mediumllm

Official pageofficial_url1

Claim history for Official page
ValueValid from → toStatusSourceConfidenceExtractor
https://qwen.ai → currentcurrentQwen — official blogT2mediumllm

Audioaudio1

Claim history for Audio
ValueValid from → toStatusSourceConfidenceExtractor
Yes → currentcurrentQwen — official blogT2mediumllm

Fine-tuning availablefine_tuning_available1

Claim history for Fine-tuning available
ValueValid from → toStatusSourceConfidenceExtractor
Yes → currentcurrentQwen — official blogT2mediumllm

Training data notestraining_data_notes1

Claim history for Training data notes
ValueValid from → toStatusSourceConfidenceExtractor
Multi-task pretraining for audio language alignment, supervised finetuning, and direct preference optimization. → currentcurrentQwen — official blogT2mediumllm

Visionvision1

Claim history for Vision
ValueValid from → toStatusSourceConfidenceExtractor
No → currentcurrentQwen — official blogT2mediumllm

Weights availableweights_available1

Claim history for Weights available
ValueValid from → toStatusSourceConfidenceExtractor
Yes → currentcurrentAI Atlas curated registry (YAML, versioned in git, every entry carries its source URL)T2highderived

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

Provenance

Attributed facts

19

Source tiers

T219

Freshest observation

3 h ago

Conflicts

None

Source documents 1

Source documents
SourceDocumentTypeTierLast observedSnapshots
Qwen — official blogqwenlm.github.io/blog/qwen2-audio newsT1· Official13 h ago1

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.

Data quality (59/100) measures how well AI Atlas knows this entity — completeness, primary-source ratio, freshness, conflicts — never how good the model is. Methodology →