Qwen3-ASR-1.7B

Name: Qwen3-ASR-1.7B
Author: Qwen

by Qwen

State-of-the-art open-source ASR for 52 languages with streaming and offline modes

320Kdl/month

1.7Bparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

Qwen/Qwen3-ASR-1.7B

Feature URI

mixpeek://transcription@v1/qwen3_asr_1b_v1

Overview

Qwen3-ASR-1.7B is Alibaba's flagship open-source speech recognition model supporting 52 languages and dialects. It combines a 300M-parameter AuT audio encoder with a Qwen3-1.7B decoder, achieving state-of-the-art performance among open-source ASR models and competing with the strongest proprietary APIs including OpenAI Whisper large v3.

On Mixpeek, Qwen3-ASR powers multilingual transcription pipelines that need broad language coverage beyond European languages. Its dual-mode architecture supports both streaming inference with 1-8 second chunks and offline processing of long recordings, making it versatile for real-time and batch workloads across 52 languages.

Architecture

AuT audio encoder (300M params, attention-encoder-decoder, 1024 hidden size) compresses audio 8x to 12.5 Hz representations. Qwen3-1.7B decoder for text generation. Dynamic flash attention window (1s-8s) enables both streaming and offline inference. Total 1.7B parameters.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so Qwen3-ASR-1.7B runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns_your_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection_id: "col_your_collection",
      documents: [
        {
          document_id: "asset-00412",
          // The model produces text, so it lands in payload. Give the
          // collection a text vector index and embed that text to make it
          // searchable rather than only filterable.
          payload: { extracted_text: modelOutput, source_key: "archive/2026/asset-00412" },
          vectors: { "text-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// universal_extractor@v1 runs google/gemini-embedding-2
// (3072-d) over a bucket, with no inference of your own.