Luma's Uni-1 just outranked Google's image AI — at half the price
Luma AI's Uni-1 ranks #1 in human preference for AI image generation, beats Google's Nano Banana on reasoning benchmarks, and costs just $0.09 per image. Free to try now.
Luma AI just launched Uni-1 — an AI image model that ranks #1 in human preference for overall image generation, style editing, and reference-based creation. It beat Google's Nano Banana (the previous leader) on reasoning benchmarks, and it's the cheapest option at high resolution: just $0.09 per image.
Not just another image generator
Most AI image tools are one-trick ponies: you type a prompt, you get an image. Uni-1 is different. It's a multimodal reasoning model — meaning it thinks about your request before creating anything.
Ask it to "age a person from childhood to old age" and it generates a coherent sequence. Give it a sketch and it turns it into a photograph. Feed it up to 9 reference images and it combines them into a new scene — keeping the style, identity, and composition you intended.
What makes Uni-1 different: Traditional image AI uses a "diffusion" process (gradually refining noise into an image). Uni-1 uses an autoregressive approach — it builds images token by token (piece by piece, like writing a sentence), with text and images sharing the same processing pipeline. This is what lets it actually reason about what you want.
The numbers that matter
Human preference rankings (Elo rating):
• #1 — Overall generation
• #1 — Style and editing
• #1 — Reference-based generation
• #2 — Text-to-image (behind Google's Nano Banana)
On RISEBench (a test for logic-based image processing — spatial reasoning, cause-and-effect, temporal sequences), Uni-1 scores the highest of any model, narrowly beating both Nano Banana 2 and GPT Image 1.5.
Pricing: cheapest at full quality
At 2K resolution (2048 pixels — sharp enough for any professional use), here's what each model costs per image:
Uni-1: $0.09 per image
Google Nano Banana 2: $0.10 per image
Google Nano Banana Pro: $0.13 per image
If you're a designer, marketer, or content creator generating hundreds of images monthly, Uni-1 saves 10-30% compared to Google's alternatives at the same quality tier.
What you can actually do with it
Designers and illustrators: Feed Uni-1 a rough sketch and it turns it into a polished photo or illustration. You control camera angle, lighting direction (golden hour, side lighting), and artistic style.
Manga and webtoon creators: Uni-1 generates full manga pages and webtoon panels — including a demo series called "Almost Human" made entirely with the model.
Brand teams: Upload up to 9 reference images (your product shots, brand colors, style guides) and Uni-1 generates new scenes that stay faithful to your visual identity.
Multilingual text rendering: Unlike most image AI that mangles text, Uni-1 renders text in multiple languages — including Chinese calligraphy and decorative scripts.
Try it now — for free
Uni-1 is available right now through the Luma Labs web interface with a free tier (30 credits/month — enough for roughly 15-30 images). Paid plans start at Pro (90 credits) and Ultra (300 credits).
API access is coming soon — Luma has a waitlist on the Uni-1 page for developers who want to integrate it into their own tools.
Supported aspect ratios: 3:1, 2:1, 16:9, 3:2, 1:1, 2:3, 9:16, 1:2, 1:3 — covering everything from Instagram stories to ultrawide banners.
Google's image AI just got a real competitor
Google's Nano Banana has dominated AI image generation for months. Uni-1 is the first model to beat it in 3 out of 4 human preference categories while costing less. For designers and creators who've been locked into Google's ecosystem, there's now a genuine alternative — and it's free to start with.
Full analysis at The Decoder. Try Uni-1 at lumalabs.ai.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments