Stability AI Launches Stability Audio 3.0 for Songs Up to 6 Minutes

Stability AI 推出音频模型 Stability Audio 3.0，可生成最长 6 分钟专业级歌曲

Stability AI launched the Stability Audio 3.0 audio generation model family with four sizes ranging from 459 million to 2.7 billion parameters; the small model targets on-device use and generates audio under 2 minutes locally, while medium and large models support full music creation beyond 6 minutes and 20 seconds.

Why it matters: HKR-H/K pass because Stability AI gives concrete duration and model-size details. HKR-R is weak: no benchmarks, licensing, pricing, or access terms are disclosed, so this sits at the featured threshold.