Adobe Unveils Firefly Image Model 5 with Layered Editing and Custom AI Creation

Adobe launched Firefly Image Model 5 at Adobe MAX 2025, introducing layered image editing capabilities and custom model creation alongside expanded audio generation tools. The public beta marks Adobe’s most significant generative AI update, targeting creative professionals who demand greater control over AI-generated content.

Firefly Image Model 5 generates images at resolutions up to 4 megapixels without upscaling, delivering enhanced photorealism and improved human rendering. The model powers a new Prompt to Edit tool that combines image generation and editing, allowing creators to describe modifications in natural language. Adobe integrated third-party models from Topaz Bloom, ElevenLabs Multilingual v2, Google’s Gemini 2.5 Flash, and Black Forest Labs’ FLUX.1 Kontext into Creative Cloud applications.

Layered Image Editing, currently in development, enables precise, context-aware compositing that maintains visual coherence across all changes made to an image. Firefly Custom Models entered private beta today and will roll out broadly next month, letting creative professionals personalize AI models by dragging and dropping images they have rights to use. Custom models remain private by default and require creator authorization for third-party access.

Image: Adobe

Audio Features Target Video Creators

Adobe introduced Generate Soundtrack and Generate Speech in public beta, addressing licensing challenges that have plagued video producers. Generate Soundtrack analyzes uploaded video clips to suggest thematically appropriate instrumental tracks that automatically sync with content timing. Creators can select from preset styles including lofi, hip-hop, classical, and EDM, or describe desired sounds in natural language. All generated tracks carry full commercial licenses.

Generate Speech combines voices from Adobe’s Firefly Speech Model with options from partner ElevenLabs, supporting text-to-speech generation in multiple languages. Creators can adjust emphasis, emotion, pacing, and individual word stress for natural voiceover delivery. Alexandru Costin, Adobe’s vice president of generative AI and Sensei, emphasized the commercial safety of both tools, noting they were “designed with creators in mind” to eliminate DMCA strikes and licensing fees.

Pricing and Enterprise Access

Firefly Image Model 5 is accessible through Adobe’s existing subscription tiers: Standard at $9.99 per month with 2,000 generative credits, Pro at $29.99 per month with 7,000 credits, and Premium at $199.99 per month with 50,000 credits. Creative Cloud All Apps subscribers receive 1,000 included credits monthly. Through December 1, Creative Cloud Pro and Firefly plan subscribers can generate unlimited images.

Adobe trained Firefly models on licensed content and provides commercial indemnification for business users. The model automatically embeds Content Credentials for transparency. Enterprise plans with custom pricing and dedicated support are available for teams exceeding 250 members.

The audio and image tools complement Adobe’s new timeline-based video editor in private beta, which enables creators to add voiceovers, soundtracks, and titles within a multitrack editing interface. The integrated approach positions Adobe to compete with standalone AI music generators and AI photo editing tools by embedding capabilities directly into familiar Creative Cloud workflows.