What's new in VIX.
June 2026
4 ships New tools
P-Video can animate portraits and performances
Pruna AIAnimate a reference image from a source video, or turn one portrait into a talking avatar with script or audio.
- P-Video Animate
- P-Video Avatar
Vidu Q2 and Q3 Turbo add fast video options
ViduGenerate short clips from text or a first frame, with Q3 adding last-frame guidance and optional native audio.
- Vidu Q2 Turbo
- Vidu Q3 Turbo
Sidechain Ducker makes room in a mix
Sober Bunny LabsFeed it a main track and a key track, then automatically lower the main audio whenever the key gets loud.
- Sidechain Ducker
Improvement
Ideogram's Design Board is more direct
Draw a box to add text or an image region, tune typography more precisely, let the AI place loose pieces, and edit the overall style from clearer controls.
1 ship Fix & polish
Generated files have cleaner names
New image, video, and audio outputs now get readable names based on the tool that made them, instead of machine-looking filenames.
7 ships New tools
Parakeet turns media into transcripts and captions
NVIDIATranscribe audio or video into clean text and a word-level subtitle track you can use for captions later.
- Parakeet-TDT 0.6B v3
Wan 2.6 Flash animates an image quickly
AlibabaTurn a first frame into a short HD video, with optional native audio when the clip needs sound baked in.
- Wan 2.6 Flash
PixVerse LipSync redubs a talking video
PixVerseConnect a source video and a voice track, then get a new video with the mouth motion matched to the audio.
- PixVerse LipSync
PixVerse V5.6 adds another video direction
PixVerseGenerate video from text or a first frame, with optional native audio, styles, and a thinking mode for harder prompts.
- PixVerse V5.6
Seedance 1.5 Pro generates video with sound
ByteDanceCreate short text-to-video or image-to-video clips with synchronized audio and an optional locked camera.
- Seedance 1.5 Pro
New feature
Start from a real workflow template
Home now offers starter workflows for common jobs like editing with a prompt, generating a video, or making a song.
New users get a guided first pass
The welcome flow now helps new creators land in the editor with a little more context and less blank-canvas guessing.
4 ships New tools
OmniHuman 1.5 makes a portrait talk
ByteDanceConnect one portrait and a voice track to generate a talking video with the same subject and voice-driven motion.
- OmniHuman 1.5
KlingAI Lip-Sync gives character clips a new voice
Kling AIRedub an existing video with new dialogue or music, then balance the original audio against the synced track.
- KlingAI Lip-Sync
xAI Text to Speech adds six studio voices
xAITurn text into speech with xAI voices, auto-detected language, and inline cues for pauses, laughs, whispers, and emphasis.
- xAI Text to Speech
Improvement
Workflow errors point to the problem
When a workflow cannot run yet, VIX now gives a clearer reason and a Show me action that jumps to the node that needs attention.
5 ships New tools
Fish Audio brings expressive voices to the canvas
Fish AudioGenerate narration, dialogue, and voiceover from text, or clone a voice from a short reference clip.
- Fish Audio S2.1 Pro
Mirelo turns prompts into sound effects
MireloDescribe the sound you need, like footsteps, ambience, impacts, or machinery, and get a clean audio track back.
- Mirelo SFX 1.5
HeyGen Avatar V makes presenter videos
HeyGenPick a digital presenter and drive it with a script or voice track to create a talking-head video.
- HeyGen Avatar V
Sync Lip Sync 2 can redub existing footage
sync.Connect a video and a new voice track, then re-sync the mouth movement while preserving the original shot.
- Sync Lip Sync 2
Improvement
Tool details open on the canvas
The same rich details from the tool library now open from nodes already in your workflow, so you can inspect a tool without leaving the canvas.
2 ships Improvement
Workflows remember how you left them
Workflow cards now show a small preview of the canvas, so it is easier to recognize a project when you come back.
Fix & polish
Fewer runs get stuck waiting
We fixed bugs that could leave some executions stuck after a worker finished, so runs recover more reliably.
7 ships New tools
SAM 3.1 can mask a moving subject
MetaType what you want selected and SAM follows it through the clip. You get a preview video and a mask video you can wire into the next edit.
Masking one image is useful. Masking a thing that moves is the harder job.
SAM 3.1 lets you write a simple prompt like "person" or "red car" and get back a tracked selection across the clip. The preview shows what it caught. The mask gives the next tool something precise to work with.
Reach for it when a subject needs to stay selected after the first frame.
- SAM 3.1 Video Segmentation
- SAM 3 Image Segmentation
Pack and Unpack Video Track keep sequences flexible
Bundle clips into one ordered video track, then pull the pieces back out later when another tool needs them separately.
- Pack Video Track
- Unpack Video Track
Improvement
Missing inputs are easier to fix
The workflow drawer now has Runs and Issues. If something is missing, you can jump straight to the node that needs attention.
The canvas is easier to start from
Empty workflows now feel less empty, new nodes land more gracefully, and the view controls are closer to where your hands already are.
Home is more useful when you come back
Fresh outputs, recent activity, starter ideas, search, and visual workflow cards make it easier to pick up where you left off.
Fix & polish
Media Hub is easier to read
Files, previews, upload state, processing badges, and output markers are easier to scan. Less panel clutter, more of the media itself.
No surprise mid-cycle plan math
During launch, changing plans will not create prorated charges or credits. You see what is due today before you confirm.
6 ships New tools
Aleph 2.0 — edit footage by describing the change
RunwayFeed it a clip, say what should be different, get the edited shot back. Video editing that takes direction in plain language.
Most AI video tools generate from scratch — you describe a world and hope. Aleph 2.0 works the other way around: you bring real footage and tell it what to change. Relight the scene from day to dusk. Swap the grade. Remove the thing in frame that shouldn't be there. Restyle the whole shot while the motion, the camera move, the timing all stay exactly as you shot them.
That's the unlock — it edits your footage instead of replacing it. The clip you spent an afternoon getting stays yours; you're just art-directing the parts you'd otherwise send to a colorist or a VFX pass.
Reach for it when:
- You love the take but hate the lighting.
- A shot needs a different look to match the rest of the cut.
- There's one distracting element you'd kill for a clean plate of.
- Aleph 2.0
LTX-2 — fast video that doesn't feel cheap
LTX StudioHigh-quality generation at a speed that lets you actually iterate instead of waiting on a single roll of the dice.
- LTX-2
Crop and Resize — the unglamorous daily image work
Crop to a region, resize to a target. The small stuff that should never force you out of your workflow into another app.
- Crop Image
- Resize Image
Six small tools that glue workflows together
Do math on any number, roll a random one, remap a range, transform and truncate text, pull a value out of structured data. Tiny on their own — powerful between two bigger tools.
- Math
- Random Number
- Remap Range
- Text Transform
- Truncate Text
- JSON Extract
New feature
See what a run costs before you press play
Every tool now carries a small bolt that shows how heavy it is on your plan's budget — on the card, before you run it. No math, no surprises.
Improvement
Favorites in the tool library
Pin the tools you reach for daily. They stay one click away instead of buried in a catalog that grows every week.
5 ships New tools
Mask Maker + Inpaint — edit exactly what you point at
Paint over a region of any image, then have an edit model change just that region and leave everything else untouched.
- Mask Maker
- Inpaint Image
Krea 2 — image generation in three sizes
KreaPick your trade-off: Turbo for fast drafts, Large when the final frame has to hold up. Same model family, three speeds.
- Krea 2 Large
- Krea 2 Medium
- Krea 2 Turbo
ElevenLabs — music and voice on the canvas
ElevenLabsEleven Music for original soundtracks, Eleven v3 for voice with genuine performance. The audio half of a workflow, handled.
- Eleven Music
- Eleven v3
Recraft Vectorize — raster in, clean vectors out
RecraftTurn a generated or hand-drawn raster into crisp, scalable vector art. Logos and icons that stay sharp at any size.
- Recraft Vectorize
New feature
The tool library, rebuilt
Browse the whole catalog with live filters and scoped search, and open any tool into a detail view with real demo media — what the model actually makes, not a stock thumbnail.
2 ships New tools
Ideogram 4.0 — the model that can actually spell
IdeogramThe one you reach for when the image needs to say something — posters, logos, thumbnails — and get the words right the first time.
Most image models treat text as decoration — squiggles that look like letters until you read them. Ideogram 4.0 treats text as text. Posters, packaging, logos, thumbnails, meme-grade typography — it lands the words clean, the first time, instead of leaving you to photoshop over alphabet soup.
But the real unlock is how you talk to it. Ideogram exposes a control surface most models don't: you can direct layout, type treatment, and composition in language — not just "a poster of X," but where the headline sits, how the type feels, what the hierarchy is. It's a genuinely new axis of control. You're not rolling dice on where the words land; you're art-directing them.
Reach for it when:
- The image has a job — a launch graphic, an album cover, a thumbnail that has to read at two inches.
- You've spent an hour fighting another model's gibberish text.
- You want the layout dialed by description, not by regenerating forty times.
- Ideogram 4.0
Fix & polish
Saving to your library is way faster
We sped up how outputs save to your library. Big files, small files — they just land, instantly.
2 ships New tools
Ramp — animate any setting over time
Draw a curve and wire it into any number a tool exposes: fade volume out, ease opacity in, drift a zoom. Automation without keyframe spreadsheets.
- Ramp
ACE-Step v1.5 Turbo — full songs in seconds
A complete track — structure, instrumentation, the works — from a text prompt, generated fast enough to keep trying.
- ACE-Step v1.5 Turbo
1 ship Improvement
Usage limits that warn before they stop you
Getting close to your plan's window? You see it coming — a heads-up first, a clear explanation if you reach it. Never a silent failure mid-run.
1 ship New tools
Eleven image models under a penny per run
A volume drop for high-throughput work — drafts, variations, b-roll frames. When you want a hundred attempts, the price of the attempt matters.
- Qwen-Image
- Qwen-Image-Edit
- SDXL
- Stable Diffusion 3
- Juggernaut Lightning Flux
- Juggernaut Pro Flux
- Kandinsky 5.0 Image Lite
- P-Image
- P-Image-Edit
- HiDream-I1 Dev
- HiDream-I1 Full
May 2026
1 ship New tools
MiMo V2.5 — one model that reads everything
XiaomiReasons across text, images, video, and audio in a single shot and answers back in text. Ask it about your footage mid-workflow.
- MiMo V2.5
1 ship Fix & polish
Smoother wiring on the canvas
Connections between tools are easier to grab, reroute, and read at a glance. Small hands-on-the-canvas stuff you feel immediately.
1 ship Improvement
Number controls you can feel
Drag any numeric setting like a dial — scrub it, nudge it by steps, or type it exactly. The canvas should feel like an instrument, not a form.
1 ship New tools
Seedance 2.0 Fast — video generation tuned for speed
ByteDanceThe latest in the Seedance line, built to return a shot quickly so generation stays part of your flow instead of interrupting it.
- Seedance 2.0 Fast
On the horizon.
Bunni Editor
Coming soonAn AI editor that works like a real editor. Annotate the cut, leave the note, point at what's wrong — Bunni reads between the lines and does the work. She's in the kitchen.
Studio Editor
Coming soonA familiar timeline editor for hands-on, multi-track composition. Picks up right where your workflows finish.
The next model drop
When a lab ships something worth your time, our job is to get it onto your canvas fast — often within days. If it's good, it lands here.