AI Research Atlas

Stable Audio 2.5

Stability AI · 10 September 2025

Stable Audio 2.5, billed as the first enterprise-grade audio model, makes three-minute tracks in under two seconds on a GPU and supports inpainting.

Post-trained with Adversarial Relativistic-Contrastive (ARC) distillation for speed; aimed at brand sound identity at scale via API, fal, Replicate, ComfyUI and on-prem licences.

Date
Wednesday, 10 September 2025
Lab
Stability AI
Kind
model
Access
closed API

Sources

  1. stability.ai/news-updates/stability-ai-introduces-stable-audio-25-the-first-audio-model-bu

This record was checked against its sources on 6 October 2026. How we check

Related