Particle.news

Adobe Expands Firefly Into Unified AI Studio as Public Betas Go Live

The rollout pairs new creative tools with partner model options under tighter training and IP safeguards.

Overview

  • Adobe opened public betas for Firefly Image Model 5, Generate Soundtrack and Generate Speech, bringing photoreal 4MP image generation with layered editing plus AI music and multilingual voiceover tools.
  • AI Video Editor, Firefly Custom Models and Firefly Creative Production are in private beta, with broader access beginning next month according to the company’s MAX announcements.
  • Project Moonlight was previewed as a conversational, agent-driven assistant for Adobe apps, with early access available via a waitlist in the coming months.
  • The ecosystem now supports third-party integrations including ElevenLabs, Google, Luma AI, OpenAI, Runway and Topaz Labs, alongside private Firefly Custom Models trained on a creator’s own style.
  • Adobe executives emphasized a conservative data approach using licensed, opt‑in sources such as Adobe Stock and contracted datasets, with contributor controls and contracts that prohibit partners from retraining on Firefly outputs.