At Adobe MAX 2025, Adobe unveiled chat-based AI assistants for Photoshop, Express, and Firefly, while also expanding support for third-party AI models from partners like Google, OpenAI, and Runway.
Adobe is rolling out conversational AI agents across its core applications. The new AI assistant in Photoshop enables users to chat with the agent, delegate creative tasks, and receive step-by-step guidance. Similar assistants are in development for Express and Firefly. Additionally, Project Moonlight—now in preview—aims to connect Adobe apps with users’ social channels for streamlined content management.
Adobe now offers direct access to a variety of external AI models within Photoshop and Firefly. Available options include Google Gemini 2.5 Flash Image, Black Forest Labs FLUX.1 Context and Flux 1.1, models from OpenAI and Runway, as well as Luma AI, Moonvalley, Pika, Ideogram, Topaz Labs for upscaling, and ElevenLabs Multilingual v2 for voiceovers.
All models are accessible through a unified interface and pricing structure. Until December 1, Creative Cloud Pro and Firefly subscribers can generate unlimited images using any of these models.
Adobe plans to integrate additional third-party models for animation, 3D asset creation, and audio design in the coming months. Users will be able to compare model outputs, set project-specific preferences, and access detailed usage analytics.
Firefly Image Model 5 and Advanced Editing Tools
The new Firefly Image Model 5 generates images natively at 4 MP resolution—eliminating the need for upscaling—and delivers highly detailed, photorealistic results. “Edit Prompts” let users describe desired changes in natural language, while layered editing enables context-aware compositing. Firefly Image Model 5 is now available as a public beta.
Firefly also introduces “Generative Soundtrack” and “Text-to-Speech”—both in public beta—for creating licensed instrumental tracks and AI-generated voiceovers. The Firefly Video Editor (private beta) offers web-based multi-track timeline editing with style presets such as claymation, anime, and 2D. For bulk workflows, Firefly Creative Production (invite-only beta) allows users to edit thousands of images at once—swapping backgrounds, adjusting colors, and cropping in bulk.
New AI Features in Photoshop, Premiere Pro, and Lightroom
Adobe is introducing several new AI-powered tools across its flagship applications. In Photoshop, users now have access to “Generative Fill” powered by partner models, “Generative Upscale” for 4K enhancement, and “Harmonization,” which automatically matches lighting, color, and tone between objects and backgrounds for seamless composites. These features are now live.
In Premiere Pro, the new “AI Object Mask” (beta) automatically identifies and isolates people and objects in video frames without manual rotoscoping. Additional masking tools like “Quick Vector Mask” make tracking during editing faster and more precise.
For Lightroom, the new “Assisted Culling” feature (beta) accelerates the process of identifying the best shots within large photo collections.
Availability: Public and Private Betas
Firefly Image Model 5, “Generative Soundtrack,” “Text-to-Speech,” and Photoshop’s “Generative Fill” (with partner models), “Generative Upscale,” and “Harmonization” are now available as public betas or fully released features.
The Firefly Video Editor, Firefly Custom Models, and Firefly Creative Production are currently in private beta and will launch next month. The Photoshop AI Assistant is accessible via a waitlist for the private beta. Project Moonlight is expected to roll out in the coming months.
“Firefly Custom Models” (private beta) lets users train their own AI models by uploading reference images to generate assets in their unique style. Through Firefly Foundry, enterprises can collaborate directly with Adobe to build custom models trained on their intellectual property, supporting image, video, audio, vector graphics, and 3D content generation.
