Adobe Max 2025: Adobe Unveils Powerful AI Assistants, Firefly Model 5, and New Audio-Video Tools

At the Adobe Max 2025 event, Adobe unveiled a host of new artificial intelligence (AI) features across its creative ecosystem, marking another major leap in its AI-driven transformation. The key announcements include AI assistants for Photoshop, Express, and Firefly, a next-generation Firefly Image Model 5, and new AI-powered tools for audio and video creation. The company is also expanding third-party integrations and offering custom Firefly model training for individual users.

Next-Generation Firefly Model 5 : Adobe Max 2025

Adobe introduced the all-new Firefly Image Model 5, which combines 4-megapixel resolution, photorealistic output, and prompt-based editing within a unified large language model (LLM). The highlight of this version is its multi-layered editing capability. Every element in a generated image is treated as an editable layer, allowing users to fine-tune specific areas or rely on AI-powered prompts for automatic adjustments.

This upgrade brings enhanced creative flexibility—users can either manually edit the assets for granular control or simply describe changes through natural language commands.

Expanded Model Integrations : Adobe Max 2025

In addition to Adobe’s in-house innovations, the company is deepening its ecosystem with third-party model integrations. The latest additions include ElevenLabs’ audio generation models and Topaz’s image generation models, providing creators with more specialised tools within the Firefly environment.

Custom Firefly Models for All Users : Adobe Max 2025

Previously exclusive to enterprise clients, custom Firefly models are now available to individual users. This feature allows creators to upload their own datasets and train models based on their personal style, ensuring visual and brand consistency across creative projects. Currently in beta, users can join the waitlist for early access.

New AI-Powered Audio and Video Tools

Adobe’s new Generate Soundtrack and Generate Speech tools take audio creation to the next level.

• Generate Soundtrack analyses an uploaded video to suggest the perfect vibe, tempo, and genre of music. Users can tweak these parameters before the tool automatically generates and syncs a studio-quality soundtrack.
• Generate Speech enables realistic voice-over generation, offering a fast and professional solution for creators.

Both tools are in beta, while Adobe’s upcoming Firefly Video Editor—which can generate, organise, trim, and sequence clips—is currently available via waitlist.

Prompt-to-Edit and Model Partnerships

The Prompt-to-Edit feature enhances editing workflows, allowing users to describe changes in plain text. The tool supports Firefly Image Model 5, Google’s Nano Banana, and Black Forest Labs’ Flux.1 Kontext model, giving creators more choice in how they design and modify content.

AI Assistants Across Adobe Platforms : Adobe Max 2025

Perhaps the most impactful reveal was Adobe’s rollout of agentic AI assistants across Photoshop, Express, and Firefly. These assistants let users perform complex edits or creative tasks through conversational prompts. Users retain full control—every AI action can be reviewed or undone at any time.

• Express AI Assistant is now available in beta.
• Photoshop AI Assistant has a waitlist open for early access.

Project Moonlight: Unifying the AI Experience

To tie everything together, Adobe previewed Project Moonlight—a framework allowing all AI assistants to collaborate seamlessly across different Adobe apps. This cross-app synergy eliminates the need for manual transitions between tools, enabling creators to maintain a continuous, AI-assisted workflow. The project is currently in private beta.

With these innovations, Adobe continues to redefine the intersection of creativity and AI, offering tools that empower creators to produce professional-quality content faster and with greater precision than ever before.