What's New
Product updates on Wavemaker — new models, tools, and features. Short notes for creators (not engineering changelog).
RSS feed-
Self-hosted GPU seconds cost less
Open-weight clips (native H3, LTX) now bill Modal time at 2× list instead of 3×. Partner APIs are unchanged.
-
Craft language for Chat, Files, and Projects
One type scale, twelve opt-in primitives, and a calmer production stage — still behind flags until each surface earns a human craft PASS.
-
Projects hold Productions — chat stays in Create
An org Project is a container with a shelf, picker, and board. Each Production is the five-view workspace. Chat is the Create-view mode inside it.
-
A new look: Wavelength
The whole app moves to a deeper studio-dark theme with a new type system and the generate gradient promoted to the product's signature.
-
Audiograms and stem split
Turn a podcast clip into a branded waveform video, or split a song into vocals and instruments — same wallet, same Files.
-
Living Studio library — characters from your films stay
People, places, and products you generate land in Studio automatically. Pin them from Cast, reuse looks, undo for 24 hours.
-
Improve, dub, and prove the cut
One Improve workspace, Dubbing v2, and a Trust Center that only indexes pages with three real demos.
-
Save a persona — emotion, locale, and voice stay with the character
Cast members now carry typed performance extras. Pick People, Places, Things, Voices, Styles, or Brands from the Project library. Licensed presenter ads stay gated until quality review.
-
A new home for every creative project
Wavemaker now starts with outcomes, keeps every production in Projects, and gives desktop and phone a calmer, faster navigation system.
-
See every step behind your generation
Chat now keeps production calm at a glance and reveals durable activity, every workflow node, retries, assets, quality checks, and safe provenance on demand.
-
Lip-sync music videos and desk shows from a track
Drop a song or a voice track plus stills — Hailuo H3 lip-syncs each slice, then we remux your audio so the cut stays on the beat.
-
One word for what you run: Workflow
Discover, Team, and Yours are all workflows now. Training stays Train a look — never a third product name.
-
Preview plays stills and clips as they land
While a video is generating, Preview assembles each still and clip into a scrubbable draft — you do not wait for the finished file.
-
Community — browse generations, rate Looks, and price your recipes
A member gallery at Community, next to Chat. Post stills and clips, rate and comment, and charge unlock or per-run credits on your own workflows and LoRAs. Create still starts in Chat.
-
Director cockpit — characters, shots, and song-length storyboards
Mint a named character in Character Studio, bind them into a film without reminting, approve shots from the Director panel, and auto-storyboard from an attached track.
-
Studios on Discover — Product Ads, Ad Remake, Music Video, Character
Four one-screen studio doors on Discover home. Fill a few fields, we mint a generation and start the existing workflow — no new catalog Try park.
-
Amateur stills, identity edit, and depth control on Krea 2
Three new Discover leaves for phone-look stills, identity-preserving edit, and optional depth ControlNet — original Wavemaker graphs, not copied community JSON.
-
Worlds — share People, Places, and Things
Publish characters, locations, and products from Community Worlds. Others can search, import or unlock them, and use them in their next generation.
-
Paused for credits? One tap now tops up and resumes your production
When a production pauses on insufficient credits, a persistent card in chat offers Add credits & continue — top up and the run resumes automatically from where it stopped.
-
Cast library — save your characters and reuse them anywhere
Save a finished generation's cast to your workspace library, then pick them from the new Cast chip in the composer — your next video reuses the exact same faces, no re-minting, no drift.
-
A calmer start — first-run that follows you, and a composer that grows with you
First-run teaching now syncs across devices, the composer starts simple and unlocks advanced pickers after your first three finished runs, and Discover leads with what you want to make.
-
Developer dashboard: per-key spend, webhook delivery log, and webhook-first docs
See exactly which API key spent what, audit every webhook delivery with retries and errors, and follow a webhook-first quickstart instead of polling. Public share pages now invite viewers to create their own.
-
Smarter ComfyUI imports — near-miss nodes swap to supported equivalents
Importing a workflow with community variants like KSampler //Inspire or Image Save now substitutes the supported equivalent automatically, with a full note of what changed.
-
Reference-sheet video, restoration tools, and HDR look
Upload 2–8 stills into one identity sheet, repair clips, and apply an HDR look — Wavemaker tools, not a model farm.
-
Batch iterators on the Nodes canvas — queue prompts × references × seeds
The new Batch panel runs a whole cross-product of variations as one queued batch, with every lane priced individually before you spend a credit.
-
Quick assets: GIFs, thumbnails, lyric videos, logo stings, meme cards
Five everyday assets, now one prompt away — export any video as a GIF, and find thumbnail, lyric-video, logo-animation, and meme-card starters in Discover.
-
Review the storyboard before production spends a credit
Turn on Review storyboard in the composer and every video pauses at the storyboard: edit scene prompts and durations, see the remaining cost, then approve to continue.
-
Team seats and client review links
Buy your plan with multiple seats — credits scale with your team — and send clients a one-tap review link to approve a deliverable or request changes, no account needed.
-
Voice memory and saved styles — your voice and look follow you
Pick a trained voice once and every narration uses it automatically, and save the exact visual style behind any ad you like as a private style you can pin next time.
-
Publish your workflow as a one-form app
Pick which inputs matter when you publish, and your workflow's Discover page becomes a simple form anyone can run — pinned to the exact version you shipped.
-
Swipe to dismiss — bottom sheets feel native on your phone
The Preview panel and every bottom sheet now close with a swipe down on the grab bar or header, with a full phone-chrome polish pass behind it.
-
Static ads — new proven ad formats, always honest
Three new high-performing ad formats — annotated feature callouts, numbered listicles, and giant-stat heroes — plus proof-honesty guarantees: testimonials, ratings, statistics, and competitor comparisons only ever render facts you supply.
-
Smarter ad QC and style variants — better ads on the first send
Image ads now self-correct when quality review finds a defect, judge aesthetics against your pinned style, and can render up to four ranked styles side by side from one brief.
-
Static ads — designed typography and adaptive ink
Ad copy now renders with real typefaces per style family, scene-aware ink color that meets contrast standards, and cleaner CTA treatments — no more one-look white-on-veil overlays.
-
Ads from a URL — paste a product page, get a full ad set
One product URL now produces a researched brief, a saved brand kit, six mixed-format image ads, and an optional UGC-style video — with zero further input.
-
More partner leaves in Discover
Hailuo, Grok, Omni, Kling first/last-frame, and ElevenLabs SFX, isolation, dialogue, and speech-to-text now have their own Try leaves.
-
Delete a generation and reclaim storage
Permanently delete a chat generation and its exclusive files to free library quota. Bound files stay until you delete the generation that uses them.
-
LTX 2.5 is the SFW cinema default
Self-hosted LTX clips now generate on 2.5 by default, with film seeds restaged from the same identity plate.
-
Characters speak in the clip
On-screen dialogue is baked into Kling and Grok clips. Mute-first still captions; narrator stays ElevenLabs.
-
Preview stays clip-by-clip after wrap
Finished video Preview is a Remotion composition you can scrub by scene — Download video still gives you the Combined MP4.
-
Images Ad ingest + Gemini Image generate
Paste a site URL or locked Headline/Body/CTA — Images Ad compiles the brief, scrapes when URL-primary, and generates the scene on Gemini 3 Pro Image at 2K.
-
Catalog Guide on Discover
Describe what you want to make — Discover returns up to three matching workflows. It never auto-runs.
-
Challenges and Wavemaker Studios
Claim credits, submit a run, and compete for jury-ratified prizes — brand Challenges plus monthly Wavemaker Studios editions.
-
Honest credit prices
Partner video, audio, and GPU meters now follow the 6.6¢ / 50% cost-plus formula — a material price cut, plus Account Usage.
-
Platform API and MCP cover the same creative verbs as chat
Agents can now iterate generations, remix workflows, engage packs, and author graphs through platform MCP — with a public OpenAPI at GET /openapi.json.
-
Edit the original production — Redo, Replace, Remove
Idle video productions can rewind in place from Chat OUTPUTS and Nodes — no new run-group, no iterate sibling for those verbs.
-
Seedance 2.5 on Wavemaker
BytePlus Seedance 2.5 — up to 30 seconds of video at 720p with native audio and multimodal references.
-
Face swap with likeness consent
Image face-swap iterate and refine flows now require an explicit likeness consent before engage — with AI-manipulated disclosure stamped on inputs.
-
Partner AI models & curated tools on Wavemaker
Named-model packaging for Kling, Seedance, Veo, WAN, and more — plus curated tools like face swap, localize, and image-to-video.
-
Video localize — five launch languages
Localize a single-speaker video into Spanish, French, German, Brazilian Portuguese, or Japanese with a proofread gate before engage.