Which AI image model should you pick?
Published
The four models at a glance
FLUX.2-klein-4B is the speed pick: a 4-billion-parameter model distilled for fast turnaround, up to 2048×2048. FLUX.2-klein-9B trades a little speed for finer detail and stronger prompt adherence at 1024². FLUX.1-schnell is the classic distilled 12B FLUX with the widest size range (1024², 1536², 2048²). Z-Image-Turbo is a 6B model tuned for photorealistic output and accurate Chinese/English text rendering in about 9 steps.
Iterating fast: FLUX.2-klein-4B
While dialing a prompt in, generation speed beats peak quality — and klein-4B is the fastest model on the site. Its model card ships a quick-start prompt that is also the simplest text-rendering test there is:
A cat holding a sign that says hello world
Try this prompt →Final renders: FLUX.2-klein-9B
Once the prompt works, switch to the 9B variant for the extra detail — its official examples lean on long, layered prompts where every clause lands. The prayer-wheel scene below specifies light direction, surface wear, shadow length and a three-color palette.
An elderly Tibetan monk in burgundy and saffron robes turns a row of copper prayer wheels along the outer wall of the Jokhang Temple in Lhasa. Late afternoon sidelight. Stone wall worn smooth. Shadow falls long across the flagstone floor. Warm ochre, aged copper, deep burgundy.
Try this prompt →The classic and the middle size: FLUX.1-schnell
schnell generates in as few as four steps and is the only model here offering 1536×1536 — handy between quick squares and heavy 2048² renders. The same quick-start prompt works on it unchanged:
A cat holding a sign that says hello world
Try this prompt →Bilingual scenes: Z-Image-Turbo
Z-Image-Turbo's headline strength is rendering complex Chinese and English text accurately, with photorealistic people. Its official quick-start example (1024², 9 steps, seed 42) is a good calibration prompt:
Young Chinese woman in red Hanfu, intricate embroidery. Impeccable makeup, red floral forehead pattern. Elaborate high bun, golden phoenix headdress, red flowers, beads. Holds round folding fan with lady, trees, bird. Neon lightning-bolt lamp (⚡️), bright yellow glow, above extended left palm. Soft-lit outdoor night background, silhouetted tiered pagoda (西安大雁塔), blurred colorful distant lights.
Try this prompt →Try them side by side
The cheapest way to decide is to run one prompt on two models and compare — every prompt card on this site, including all of the gallery, carries the model and size used so you can reproduce it in one click and then swap the model in the generator.