April 2026 Highlights: Maker Mode with generous daily limits · train your own LoRAs · the new Video Agent · smarter usage-based pricing · and a massive wave of new image, video, 3D, audio, and avatar models — each with prompt ideas to get you started.
April was one of our biggest months ever. Here's everything new.
✨ Maker Mode
Maker Mode gives you generous daily usage on a curated set of models, with clear per-model daily limits so heavy creators get predictable access without burning through credits. You'll see your remaining daily generations right in the model picker.
🧪 Train Your Own LoRAs
You can now train custom LoRA models directly in Pixio. Upload a set of images of a character, product, or style, train, and then generate with your fine-tune — including on FLUX 2. It's the fastest way to get a consistent look or character across everything you make.
🎬 The Video Agent
The new Video Agent turns an idea into a multi-shot video. Describe what you want and it helps plan the story, pick models, and generate it segment by segment — managing references so characters and scenes stay consistent across shots.
💸 Smarter, Usage-Based Pricing
Credits now scale with the options you actually choose — resolution, duration, steps, and quality. Pick a quick draft and pay less; go to 4K and you'll see the cost up front. Every model's controls show how they affect price before you generate.
🖼️ New Image Models
FLUX 2
The latest from Black Forest Labs lands on Pixio with sharper detail, stronger prompt adherence, and excellent typography. Great for product shots, posters, and photoreal scenes.
Try this prompt: "Editorial product photograph of a matte-black ceramic coffee mug on a wet slate countertop, morning window light, shallow depth of field, wisps of steam, minimalist Scandinavian kitchen blurred in the background, 50mm, ultra sharp, photorealistic."
GPT Image 2
OpenAI's newest image model, now with Maker Mode support. It's exceptional at following complex instructions and rendering clean, correctly-spelled text — ideal for thumbnails, ads, and infographics.
Try this prompt: "A bold YouTube thumbnail, 16:9. A surprised astronaut floating in a neon-lit space station, the words 'ZERO GRAVITY' in large punchy yellow sans-serif text across the top, high contrast, vivid colors, cinematic lighting."
WAN 2.7 (Image)
New WAN 2.7 image models for crisp, stylized, and photoreal generations — a great all-rounder.
🎥 New & Updated Video Models
A big month for video. Highlights:
- Kling v3 — new 4K models plus motion control and edit
- Seedance 2 — the full family: Seedance 2, Seedance 2 Maker, Max, and VIP tiers
- LTX 2.3 — reference-to-video, so you can drive a clip from a still
- WAN 2.7 — Pro edit and reference-to-video
- Veo 3.1 Lite — Google's fast, high-quality video model
- Vidu Q3 — reference-to-video
- Atlas Cloud — new image-to-video models
- Minimax 2.6 — updated video generation
- Topaz — video upscaler options to take clips up to crisp HD/4K
Try this prompt (Kling v3 / Veo 3.1 Lite): "Cinematic drone shot flying low over misty pine forest at sunrise, golden light breaking through fog, slow forward push, volumetric god rays, film grain, 24fps cinematic look."
🧊 New 3D Models
- Meshy 6 — now with multi-image input for more accurate, detailed 3D models from several reference angles
- Tripo 3.1 & P1 — faster, cleaner geometry for text- and image-to-3D
Try this prompt (Meshy 6 / Tripo): "A stylized low-poly treasure chest, weathered oak planks with iron banding and a brass lock, game-ready asset, clean topology, PBR textures."
🎵 New Audio & Voice
- Mureka — a new music generation model for full tracks and instrumentals
- Cohere & xAI — new transcription options for fast, accurate speech-to-text
Try this prompt (Mureka): "Upbeat lo-fi hip-hop instrumental, warm vinyl crackle, mellow Rhodes piano, soft boom-bap drums, 80 BPM, relaxed and nostalgic."
🗣️ New Avatars & Lip Sync
- Argil Avatars — generate talking presenter avatars
- MagiHuman — expressive, natural human animation
- Happy Horse — added via Replicate for lively character motion
- Phota & Pruna — new model options for portraits and stylization
🐛 Fixes
- Fixed Mureka instrumental generation and song uploads
- Pixio Music now costs a small flat credit amount per generation to keep it fast and fair for everyone
That's April. May brings the public Pixio API, a reworked Boards & Canvas, and even more models — see you in the next one.