ClosedAlibaba/10 trustVerified

Qwen-Audio-3.1

Name: Qwen-Audio-3.1
Rating: 7

Five-model audio family with a full-duplex Realtime model built for tool-calling voice agents.

Visit official page →

About

Qwen-Audio-3.1 is Alibaba's five-model audio stack, released September 23, 2026: a base ASR model, ASR-Next with speaker identification and emotion detection, TTS with cross-language voice transfer, TTS-Next that generates voice, sound effects, and background audio in one pass, and a full-duplex Realtime model for voice agents that can call tools mid-conversation.

The Realtime model offers 262K tokens of context, with audio input at $6.4 per million tokens, text input at $0.8, and combined text-and-audio output at $24, while plain text output is free. Alibaba cut prices sharply at launch: about 70% off TTS, 85% off Realtime, and up to 95% off ASR. None of it ships as open weights; it is a managed API on Qwen Cloud.

Specs

Context window

262,144 tokens (Realtime); 245K max input, 16K max output

Pricing

Realtime: $6.4/M audio input, $0.8/M text input, $24/M combined text-and-audio output (text output free); steep cuts vs prior generation

Free quota

API access

Qwen Cloud API (managed), WebSocket for Realtime

Weights

Closed — API only

Released

September 23, 2026

Benchmarks

No vendor benchmark table published; positioned on price (TTS -70%, Realtime -85%, ASR -95% vs prior generation)

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

voice-agentscall-centerstranscriptionmultilingual-dubbing

Trust score — 7/10

/10 trustVerified

Scored 7/10 (verified tier, metrics-v1). Evidence: current generation with full specs and pricing published by Alibaba on its official blog and Qwen Cloud. Closed weights, managed API only. Release facts corroborated across multiple independent outlets; docked for no vendor benchmark table.

How the trust score is built

Full methodology →

Adoption · stars/downloads, log-scaled vs peers: 6.0/10
Maintenance · recency, releases, issue load: 9.0/10
Openness · license + docs: 3.0/10
Manual review · sanity read + red-flag check: 8.0/10

· Adoption 6: managed API on Qwen Cloud; aggressive price cuts aimed at voice-product builders
· Maintenance 9: released Sep 23, 2026; current generation
· Openness 3: proprietary managed API, no open weights
· Manual 8: release facts (Sep 23 2026; five models ASR/ASR-Next/TTS/TTS-Next/Realtime; 262K Realtime context; $6.4/$0.8/$24 pricing; 70-95% price cuts) corroborated across The Decoder, aidailypost, and tpsreport; docked for no vendor benchmark table