Qwen-Audio-3.0-Realtime Released
千问大模型 · wechat · 2026-07-15
Alibaba has officially released Qwen-Audio-3.0-Realtime, focusing on real-time voice interaction that is both fast and smart, achieving millisecond-level responses while retaining reasoning capabilities. The official release includes Plus and Flash versions, optimized for stronger reasoning and lower latency, respectively.
The post highlights four major capability upgrades:
- IQ/Reasoning: Improved performance on benchmarks like VoiceBench and AudioMultiChallenge, with minimal degradation even under conversational inputs.
- Agent Tool Calling: Can autonomously trigger external tools without explicit instructions. Results feed into conversation memory and can be accessed via FunctionCall, MCP, API, and knowledge bases.
- Empathetic Expression: Adjusts tone, pacing, and pitch based on context, enhancing naturalness through paralinguistic signals like laughter, sighs, and hesitations.
- Duplex Interaction: Supports listening while speaking, seamless interruptions, and multi-speaker scenarios. Voice profiles can also be locked using audioprompt.
The official release attributes these capabilities to On-Policy Distillation and multi-teacher distillation: reasoning skills from text LLMs are distilled into the speech model, while different teachers are used to enhance conversational preferences, general Q&A, Agent calling, and audio understanding.
Related event: Alibaba Releases Real-time Voice Model Qwen-Audio-3.0-Realtime(2 posts)→
More from coding & agent
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11
- Astra storyboards plus Minimax H3 per-shot generation boost video success rates — Hailuo_AI · 2026-09-11
- Codex tip: use Sol with Astra and Luna sub-agents to save usage — pvncher · 2026-09-11
- agents-best-practices: a provider-neutral Agent Skill for designing and auditing agentic harnesses — tom_doerr · 2026-09-11
- Cognition's SWE-2 uses a KKT duality argument in RL to shift the effort Pareto curve — YouJiacheng · 2026-09-11
- First-ever Three.js Conference lands in Paris, with a panel on AI-shortened design workflows — OdinLovis · 2026-09-11