September 2025 Highlights: SoundCraft (Stable Audio 2.5) adds inpainting + audio‑to‑audio · VeedCraft lip‑sync (image+audio) · VidCraft HD → Kling 2.5 · ByteCraft → Seedance V4 · Prompt Optimizer · app‑wide tooltips · refreshed docs.
📚 Up-To-Date Docs and Useful Tooltips
Quality of Life First: Up-to-date docs, helpful tooltips, and faster onboarding across the platform.
Docs are refreshed with practical prompting guides and step‑by‑step instructions for every tool. In the menu, all tools now include tooltips with links to docs, supported modalities, and the underlying model name.
✍️ New and Improved Prompt Optimizer
Model‑aware prompts tailored to your requests. The new Optimizer understands generation type — Images, Videos, 3D — and the selected model to suggest the right parameters and phrasing. Save prompts, favorite them, and sort your library.
- Model‑aware tips
- Reusable presets
- Cross‑device sync
🖼️ New Image Edit Model
ByteCraft is now powered by ByteDance Seedance V4 — our best image editing model yet for consistency and control.
Prompt: "Add baroque style furniture, statues, and make this feel like someone lives here"
Before
After
Features
- Use multiple images to create scenes from references
- Native 4K image editing and generation
- Most accurate and consistent editing on the market with text cohesion, depth understanding, and character consistency
Use cases
- Real estate photography
- Product photography
- Consistent character generation
- Generating scenes from multiple angles
🎬 New Video Edit Models
WanCraft adds:
- Edit — make scene‑wide changes across the whole video
- Replace — swap characters or objects mid‑clip
- Animate — drive character animation from a reference video
WanCraft Replace example
Input Video
![]
Input Image
Result
🗣️ New Voice Models
ToneCraft powered by Minimax Voice
Latest speech models from Minimax, Voice 2 and Voice 2.5. Clone voices and generate short and long format speech using your cloned voices.
VoiceCraft updated
VoiceCraft, powered by ElevenLabs now supports ElevenLabs v3, has more text to speech controls and a community voice library of 3,000+ voices.
👄 Major Upgrades to Lip Syncing
VeedCraft added
Powered by Veed Fabric 1, this model takes an audio track and a portrait image to create a realistic talking head video with arm movement and naturalism.
LipSync Max updated
Updated with the latest ByteDance OmniHuman 1.5 model. Audio‑driven talking head videos now support more hand and body movement for enhanced realism.
🆕 New and Updated Models
- VidCraft HD — Kling 2.5
First reasoning video model - ComposeCraft — Minimax 1.5
Improved music gen - SoundCraft — Stable Audio 2.5
Audio gen & editing - WanCraft — Wan 2.5
New video + audio model - SongCraft — Suno v5
New ultra music gen model - FxCraft — PixVerse v5
Updated to include PixVerse v5 - EffectsCraft — ElevenLabs
Cinematic sound effects - ObjectCraft — Meshy 6 Preview
More detailed 3D models