Skip to content

Qwen3-ASR 0.6B

Current status and review date

Decision snapshot

  • Product role: Benchmark-qualified English one-shot STT challenger.
  • Selected or best-qualified configuration: Official Qwen base image with vLLM 0.14.0 and the fail-closed official output parser, 16,384-token maximum model length, four maximum sequences, and normalized WAV input.
  • Measured hardware: One RTX 5090 on Primary Node through an isolated endpoint; Qwen2.5-Omni remained co-resident and the RTX PRO 6000 was protected.
  • Evidence: 3.621% primary-human micro-WER, 113.58 ms sequential p95, 137.36 ms concurrency-four p95, and zero final failures or repetitions.
  • Decision: challenger, no-promotion; it met the declared non-inferiority margin, but the retained record kept Parakeet routed.
  • Important limitation: Only English, non-streaming, one-shot transcription was qualified; multilingual behavior, streaming, translation, diarization, and VAD were not tested.
  • Review dates: Retained evidence cutoff: 2026-07-28. Dossier-format review: 2026-08-31.

Review narrative

2026-07-28 — Runtime adaptation and smoke qualification

The first launch failed because the pinned runtime parsed CUDA_VISIBLE_DEVICES as an integer, and the next launch rejected the custom qwen3_asr architecture before the official package registered it. The managed image retained Docker's device reservation, exposed ordinal 0 inside the restricted container, imported qwen_asr before vLLM, and applied the official output parser fail-closed. The final endpoint returned valid one-shot transcripts rather than Qwen's internal provider envelope.

2026-07-28 — Multi-sample corpus qualification

Qwen3-ASR completed the normalized English corpus sequentially and at concurrency four. Its 3.621% primary-human micro-WER met the declared non-inferiority margin, with lower p95 latency than the retained Parakeet baseline in this campaign. The evidence qualified a challenger, not a route change.

Immutable identity

Model

  • Checkpoint: Qwen/Qwen3-ASR-0.6B.
  • Revision: 5eb144179a02acc5e5ba31e748d22b0cf3e303b0.

Runtime

  • Official base image digest: sha256:fb75b775f089e06e5a1aaebffd421e37505cc630d50c86d889d95ffa45a7e16a.
  • Derived image tag: anvil-qwen3-asr:official-vllm-0.14.0.
  • Derived image ID: sha256:8bc0588b53f601ed2a8b9da23a89fa3d90f27bef70ea73d2efde2ba8f93ccc1c.
  • Engine: vLLM 0.14.0 with the official Qwen output parser.

Tested hardware and topology

Isolated candidate lane

  • Host label: Primary Node.
  • Measured device: one RTX 5090.
  • Candidate endpoint: isolated, loopback-only serve.
  • Co-resident workload: Qwen2.5-Omni remained running.
  • Protected device: the RTX PRO 6000 was not used by the candidate.

Other accelerator products and multi-GPU execution were not tested.

Engine, quantization, KV, context, and concurrency recipe

Qualified recipe

  • vLLM 0.14.0 with qwen_asr imported before server startup.
  • Official qwen_asr.parse_asr_output applied before returning the OpenAI transcription response.
  • Maximum model length: 16,384 tokens.
  • Maximum sequences: four.
  • Corpus schedules: three sequential repetitions and one concurrency-four run.
  • Weight quantization and KV dtype: Not recorded in the retained dossier evidence.

The public experiment overlay and Compose service pin the reconstructable candidate configuration.

Evidence by measurement class

Functional

  • One-shot transcription contract: pass after the fail-closed parser patch.
  • Final failures and repetitions: zero.

Quality

  • Primary-human micro-WER: 3.621%.
  • Declared non-inferiority margin versus the retained baseline: pass.

Capacity and latency

  • Sequential p50/p95: 67.40/113.58 ms.
  • Concurrency-four p50/p95: 83.98/137.36 ms.
  • Concurrency-four throughput: 47.25 requests/s.

Evidence

The dated finding links corpus provenance, sequential and concurrency-four results, startup identity, and the friction log.

Decision and promotion state

Qualified challenger

  • Benchmark-qualified for the recorded English, non-streaming, one-shot STT contract.
  • Met the declared non-inferiority margin.

Not authorized

  • no-promotion.
  • The retained decision kept Parakeet routed and does not imply current live state after the evidence cutoff.

Failures and gotchas

Resolved runtime failures

  • The first launch failed on GPU-UUID parsing; the managed container retained the Docker device reservation and exposed ordinal 0 internally.
  • The second launch rejected the custom architecture; importing qwen_asr before vLLM registered it.
  • The default 65,536-token context did not fit beside the protected workload; the qualified one-shot recipe pins 16,384.
  • Raw provider envelopes were rejected until the official parser produced the transcript contract.

Unqualified behaviors

Multilingual recognition, streaming latency, translation, diarization, and VAD were not tested.

Dated run history