Picsart Media Studio is now available inside Claude, so users can turn their clips into finished videos without leaving the chat. Users connect it, sign in with their Picsart account, and describe the video in plain words. The finished file is saved to their Picsart Drive.
Claude can now edit videos when Picsart Media Studio is connected, which it could not do alone.
Users can add captions, cuts, transitions, and a vertical 9:16 resize in the same chat.
Each finished video is shown in a player inside the chat and saved to Picsart Drive.
The tool is meant for people who already use Claude and want short videos made from their own clips.
Picsart has added Seedance 2.5 Flash, a faster and cheaper version of ByteDance's Seedance 2.5 video model, to its AI Playground. It makes clips from 4 to 30 seconds with sound, at 480p or 720p. The model is meant for quick testing before a final video is made.
Flash makes video faster and cheaper than the full Seedance 2.5 model.
It can create clips up to 30 seconds long with sound included.
Output is capped at 720p, and the reference limit is lower at 20 files.
Picsart explained how its AI Agents now talk with users by voice, using OpenAI's GPT-Live. The creative work keeps running in the background even after the call ends. Picsart said the feature arrived in September.
Users can talk to the Picsart agent, which replies by voice while tasks keep running.
Voice conversation and the work behind it are split, so the agent can plan and use tools while talking.
Audio goes directly between the browser and OpenAI, which keeps replies fast.
Voice, typing, and clicks share one thread, so a project can be started by voice and finished by typing.
BytePlus's AI video model Seedance 2.5 can now create 4K video inside Picsart. It can produce a single scene up to 30 seconds long in 4K.
You can make low-resolution 480p drafts first and convert only the one you like to 4K.
The 4K version reuses the same prompt, references, and seed, so the framing and movement stay the same.
According to BytePlus's estimate, about 20 previews plus one final render cost up to 77% less than 20 full-resolution takes, and run nearly twice as fast.
MiniMax's new image model, H3 Image, is now available in Picsart's AI Playground. You can add up to 16 reference photos and a text prompt to create scenes at up to 4K resolution.
Users can add up to 16 reference images along with a prompt.
Results can be exported in up to 4K and edited further with the same model.
It is MiniMax's first model built specifically for still images.
The article shows a candle brand making a winter campaign set from six existing photos.
ByteDance's new image model, Seedream 5.5, is now available in Picsart. It keeps untouched parts of an image the same across many edits and can create transparent backgrounds. The company says photos look more realistic and small text stays readable.
ByteDance Seed announced Seedream 5.5 on November 9, 2026.
A demo showed one image edited 100 times in a row with its unchanged parts preserved.
The model outputs transparent-background PNG images without a separate cutout step.
Light, skin and materials look closer to a photo, and small text stays readable.
Google released Gemini Omni 1.2 Flash, its AI video model, in November 2026. It makes clips longer than the previous 1.1 Flash version and sticks more closely to the reference images you provide. Picsart lets users run both versions side by side.
Clip length rises from 10 seconds to 15 seconds.
Faces, garments, packaging text and brand colors stay more consistent across camera angles.
Users can change a product, background, lighting or aspect ratio through conversation without redoing the whole scene.
Voices in scenes with several speakers stay distinct, and lip-sync looks more natural.
AI relighting changes the light in a video after it has been shot, including its direction, color, time of day and weather. There is no need to film the scene again. Picsart's blog explains how to do this step by step.
Picsart Flow's Relight template shows several moods on one canvas
The same person and scene get different light, such as a festival at sunset or police lights
Photos get their light from a text prompt before they are turned into moving video
Light is adjusted with three controls: light, time of day and weather
Picsart has added Google's Gemini Omni 1.2 Flash AI video model to its AI Playground. Users can draft clips quickly at 360p and render the best take in 4K. The model is now available to more than 100 million Picsart creators.
A draft mode creates quick, low-resolution test versions.
Final videos can be rendered up to 4K.
A single clip can now run up to 15 seconds, up from 10.
Products, characters, and logos stay more consistent when reference images or videos are used.
Picsart has added OpenAI's latest image model, GPT Image 2.6, to its AI tools. The model comes in two forms, Flare for quick work and Sunburst for detailed work. It generates images 50% faster than the previous version and keeps characters, products, and brand details consistent across edits.
GPT Image 2.6 was released on November 4, 2026.
Flare can produce an image in as little as six seconds at low quality.
Characters, products, and brand details stay consistent across editing rounds.
New features include animated GIFs, markup-guided edits, and SVG output.