Skip to content
Model Pareto

Image generation models ranked by quality, price and speed

BenchmarkText-to-imageImage editing

Quality vs price

6 models on the frontier — nothing else is both cheaper and better.

Lower price is better. Pareto frontier: Z-Image Turbo, FLUX 3 Image Klein, Muse Image, MAI-Image-2.6-Flash, Nano Banana 2.1, GPT Image 2.6 Sunburst. 3 models are not plotted: MAI-Image-2.7-Preview, Qwen-Image-2.1, Ideogram 4.5.

3 models with no price data — shown in the strip at the left edge

MAI-Image-2.7-Preview, Qwen-Image-2.1, Ideogram 4.5

  • Best-value frontier (nothing is both cheaper and better) · strong + medium evidence; weak shown but excluded
  • Evidencestrong → weak
  • Estimated from other categories

Ranking · 40 models

Price $/image · Speed s/image

Image generation ranking. Column headers sort the table. Expand a row to see per-benchmark scores.
Select to compareSourceDetails
198.3±3.5$0.21est.~66s2/6 benchmarks measuredVerified
294.8±5.2$0.21est.~66s2/6 benchmarks measuredVerified
391.7±3.2$0.21108s4/6 benchmarks measuredVerified
487.6±8.8——1/6 benchmarks measuredVerified
583.9±3.2$0.2156s4/6 benchmarks measuredVerified
676.0±3.2$0.21118s4/6 benchmarks measuredVerified
772.0±8.8$0.060est.~23s1/6 benchmarks measuredVerified
871.3±3.9$0.034est.~6.8s4/6 benchmarks measuredVerified
968.9±3.2$0.039est.~17s4/6 benchmarks measuredVerified
1067.3±3.2$0.060est.~13s4/6 benchmarks measuredVerified
1160.1±5.2$0.048est.~19s2/6 benchmarks measuredVerified
1259.3±4.3$0.11est.~37s2/6 benchmarks measuredVerified
1358.7±4.3$0.019est.~9.5s2/6 benchmarks measuredVerified
14
Muse ImageFrontier
58.0±3.2$0.01014s4/6 benchmarks measuredVerified
1555.7±3.2$0.0678.9s4/6 benchmarks measuredVerified
1654.6±3.2$0.04828s4/6 benchmarks measuredVerified
1752.7±3.2$0.09068s4/6 benchmarks measuredVerified
1852.5±3.2$0.1317s4/6 benchmarks measuredVerified
2048.5±3.4$0.04011s3/6 benchmarks measuredVerified
2147.1±3.5$0.0505.4s3/6 benchmarks measuredVerified
2244.6±4.3——4/6 benchmarks measuredVerified
2344.3±8.1$0.10est.~24s1/6 benchmarks measuredVerified
2443.0±4.3$0.0306.2s2/6 benchmarks measuredVerified
2542.5±3.2$0.0343.1s4/6 benchmarks measuredVerified
2642.4±5.2$0.032est.~12s2/6 benchmarks measuredVerified
2742.2±5.2$0.040est.~14s2/6 benchmarks measuredVerified
28
FLUX 3 Image KleinFrontierOpen
40.7±5.2$0.009est.~5.0s2/6 benchmarks measuredVerified
2939.4±4.2——3/6 benchmarks measuredVerified
3037.6±5.2$0.016est.~6.5s2/6 benchmarks measuredVerified
3137.3±3.2$0.10est.~29s4/6 benchmarks measuredVerified
3235.3±4.3$0.024est.~9.4s2/6 benchmarks measuredVerified
3335.2±4.4$0.040est.~14s2/6 benchmarks measuredVerified
3433.6±4.3$0.10est.~29s2/6 benchmarks measuredVerified
3530.5±3.2$0.035est.~15s4/6 benchmarks measuredVerified
3827.2±5.6$0.030est.~11s1/6 benchmarks measuredVerified
3926.9±4.3$0.06034s2/6 benchmarks measuredVerified
4125.9±5.6$0.035est.~13s1/6 benchmarks measuredVerified
4424.5±4.7$0.21est.~53s2/6 benchmarks measuredVerified
4621.3±3.3$0.0407.3s5/6 benchmarks measuredMixed
48
Z-Image TurboFrontierOpen
16.6±3.9$0.0051.9s4/6 benchmarks measuredMixed

About image generation

Text-to-image, image editing, and text rendering. Quality is a weighted average of the benchmarks below, each rescaled so 100 = the best tracked model. Benchmarks without a published score are estimated from related ones (shown dashed). Dot fill shows the evidence: solid = strong, tinted = medium, empty ring = weak (few benchmarks measured, lab-reported only or provisional). Methodology

Day-0 (model cards) Third-party

Top quality
GPT Image 2.6 Sunburst
98.3 / 100
Cheapest on the frontier
Z-Image Turbo
16.6 quality · $0.005
Newest
GPT Image 2.6 Sunburst
Nov 4, 2026
Fully verified
94%
of 50 models