変更履歴

Two new video models landed this week. One goes hard on quality. One goes hard on efficiency. Together they cover both ends of the creative spectrum.

🎬 Seedance 2 & Seedance 2 Fast

ByteDance's latest video engine, built for creators who need industrial-grade consistency and cinematic motion.

• Smooth camera movement: pans, tracks, and dolly shots stay stable
• Character consistency: the same subject across every frame
• Seedance 2 Fast: same engine, optimized for speed

⚡ Veo 3.1 Lite

Google's newest efficiency-tuned model, built for creators producing at scale. When you need a lot of video and the budget has to work hard, this is the one.

Try it now

Grok Imagine now supports any duration between 6 and 30 seconds, billed per second. Need 8? Pay for 8. No more paying for a fixed 15 when your clip needs 9.

⚡ What's new

• Flexible duration: 6s, 10s, 15s, 20s, 30s, anywhere in between
• Per-second billing: no round-ups, no forced minimums
• Same quality at every length: a 30-second clip renders at the same fidelity as a 6-second one
• Ideal for short-form: social cuts, transitions, and loops now priced fairly

Try it now

A big drop this week. Three new models, a resolution bump for Seedance 1.5 Pro, and a new moderation layer on uploads.

🎨 Qwen 2 for image generation

Alibaba's latest, strong on Asian aesthetics, character work, and detailed layered compositions. A solid choice when Western-trained models start to wobble on these subjects.

🎬 Runway Gen-4 Turbo

Runway's speed-tuned Gen-4 variant. Same visual ceiling as the full model, short enough round trip to actually iterate on.

📺 Seedance 1.5 Pro, now 1080p

Full 1080p output, with pricing refined accordingly. If you've been on Seedance 1.5 Pro, you'll see the new resolution immediately.

🛡️ NSFW input moderation

We now screen prompts and reference images at upload time. Most people won't notice it. If something doesn't pass, the message tells you what and why — we're trying not to be a black box about it.

Try the new models

Feed Video Extender a short clip and it generates more video continuing from where yours ends. Same subject, same lighting, same scene logic.

🎞️ What you can do

• Continue any video: upload a short clip, get more footage
• Motion consistency: camera, subject, and scene dynamics carry through
• Prompt-guided extension: describe where the story should go next
• Perfect for: B-roll, social cuts, ad extensions, longer narratives from short sources

Start with a clip you'd be happy to double, not a 3-second one you want to stretch to 60.

Try it now

Need to turn a square into a 16:9 banner? Or stretch a portrait into a wide header? Image Extender does exactly that, powered by Ideogram V3 Reframe.

🖼️ What you can do

• Expand any canvas: widen, heighten, or reframe in one click
• Seamless continuation: AI extends content to match your original style
• Any aspect ratio: portrait to landscape, square to 16:9, whatever you need
• Subject stays put: the original composition is untouched

Try it now

The model picker now tells you how many credits each model needs before you click generate. If you don't have enough, the model shows a lock icon instead of silently failing.

✨ What changed

• Credit cost shown inline: each model displays its cost right in the picker
• Lock icon for insufficient balance: no more clicks that go nowhere
• Upgrade prompt on tap: tap the lock to see how to unlock
• No changes for Pro and Lifetime: full access as always

This was the single biggest source of "why didn't that work" in our inbox. Fixing it felt worth writing down.

The gap between Free and Standard was too wide for a lot of you. Say hi to Starter — the most affordable way to go paid, priced at $9/month.

💰 What Starter gives you

• $9 / month: the most affordable paid tier
• Core tools: full access to essential image and video generation
• Monthly credits: enough for regular creation
• More concurrency: run more generations in parallel than Free
• Month-to-month: cancel anytime, no contract

Existing subscribers are untouched. Standard and Pro stay exactly as they were.

See the plans

Give it a still portrait and a short reference clip. It transfers the performance onto the portrait. Facial expression, head movement, micro-gestures, all without a single blendshape or rig.

🎭 What you can do with it

• Portrait-to-video: one still + one motion clip = animated character
• Expression transfer: capture the energy of a reference performance
• Zero 3D work: no rigging, no modeling, two inputs and you're done
• Powered by Kling: the same engine behind our flagship video models

Try it now

Google's Nano Banana 2 is now available for text-to-image and image editing. If the previous version let you down on legible text or multi-image workflows, this is the upgrade you were waiting for.

🎨 What Nano Banana 2 brings

• Readable text inside images: posters, covers, labels, typography
• Multi-image workflows: blend character and scene references cleanly
• Subject consistency: the same character holds up across edits
• Design-quality output: strong enough for client-facing work

Try it now

Kuaishou's newest video model is live in Mojomake. A meaningful jump from 2.x, especially if you've been burned by characters morphing mid-shot.

🎬 What Kling 3.0 does well

• Character consistency: the same face, outfit, and build across every shot
• Camera direction: pan, track, dolly actually follow your prompt
• Dialogue scenes: lip-sync works, which wasn't reliable before
• Standard and Pro modes: draft with Standard, deliver with Pro

Try it now

Mojomake now works in 7 languages. Create with AI in the language you think in.

🌐 Languages we now support

• 🇺🇸 English · 🇨🇳 简体中文 (already shipped)
• 🇩🇪 Deutsch
• 🇪🇸 Español
• 🇫🇷 Français
• 🇯🇵 日本語
• 🇵🇹 Português

What's translated

The entire workspace, not just the marketing pages. Every tool, setting, pricing detail, email notification, and onboarding step speaks your language.

Switch from the language picker in the header or footer. If a translation reads awkwardly somewhere, tell us — we'll fix it.

Mojo Your Ideas, Make Magic Happen.

Mojomake is officially live. One workspace for every kind of AI-generated visual content, connected to the models we think are actually worth using.

What's in the launch

• Text-to-Video: cinematic video from a single prompt
• Text-to-Image: high-fidelity image generation
• Image-to-Video: turn stills into motion
• Effects & Transforms: a creative toolkit that keeps growing
• Subscription & credits: flexible plans for hobbyists through professionals

Start creating