Image generation models ranked by quality, price and speed
/
Quality vs price
5 models on the frontier — nothing else is both cheaper and better.
Labels
Lower price is better. Pareto frontier: Z-Image Turbo, Muse Image, MAI-Image-2.6-Flash, MAI-Image-2.6, GPT Image 2.5 Sunburst. 1 models are not plotted: Qwen-Image-2.1.
1 model with no price data — shown in the strip at the left edge
Qwen-Image-2.1
- Best-value frontier (nothing is both cheaper and better)
- Evidencestrong → weak
- Estimated from other categories
- Hover a dot for details · click a lab to highlight it
Ranking · 38 models
Price $/image · Speed s/image
| Select to compare | Source | Details | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | GPT Image 2.5 SunburstFrontier | OpenAI | 99.5±1.9 | $0.21 | 108s | 4/6 benchmarks measured | Verified | |||
| 2 | OpenAI | 91.7±3.2 | $0.21 | 56s | 4/6 benchmarks measured | Verified | ||||
| 3 | OpenAI | 82.6±3.2 | $0.21 | 118s | 4/6 benchmarks measured | Verified | ||||
| 4 | MAI-Image-2.6Frontier | Microsoft AI | 75.3±3.2 | $0.039 | est.~17s | 4/6 benchmarks measured | Verified | |||
| 5 | SpaceXAI (xAI) | 70.7±3.2 | $0.060 | est.~13s | 4/6 benchmarks measured | Verified | ||||
| 6 | MAI-Image-2.6-FlashFrontier | Microsoft AI | 65.7±4.3 | $0.019 | est.~9.5s | 2/6 benchmarks measured | Verified | |||
| 7 | Muse ImageFrontier | Meta | 63.8±3.2 | $0.010 | 14s | 4/6 benchmarks measured | Verified | |||
| 8 | Microsoft AI | 61.8±4.3 | $0.11 | est.~37s | 2/6 benchmarks measured | Verified | ||||
| 9 | Google DeepMind | 61.8±3.2 | $0.067 | 8.9s | 4/6 benchmarks measured | Verified | ||||
| 10 | Microsoft AI | 60.9±3.2 | $0.048 | 28s | 4/6 benchmarks measured | Verified | ||||
| 11 | ByteDance Seed | 58.0±3.2 | $0.090 | 68s | 4/6 benchmarks measured | Verified | ||||
| 12 | Google DeepMind | 58.0±3.2 | $0.13 | 17s | 4/6 benchmarks measured | Verified | ||||
| 13 | OpenAI | 57.0±3.2 | $0.13 | 28s | 4/6 benchmarks measured | Verified | ||||
| 14 | Alibaba Qwen | 54.3±3.5 | $0.040 | 11s | 3/6 benchmarks measured | Verified | ||||
| 15 | SpaceXAI (xAI) | 51.9±3.5 | $0.050 | 5.4s | 3/6 benchmarks measured | Verified | ||||
| 16 | Qwen-Image-2.1Open | Alibaba Qwen | 50.6±4.3 | — | — | 3/6 benchmarks measured | Verified | |||
| 17 | HiDream.ai | 49.5±8.2 | $0.10 | est.~24s | 1/6 benchmarks measured | Verified | ||||
| 18 | Alibaba Qwen | 48.1±4.3 | $0.030 | 6.2s | 2/6 benchmarks measured | Verified | ||||
| 19 | Google DeepMind | 47.9±3.2 | $0.034 | 3.1s | 4/6 benchmarks measured | Verified | ||||
| 20 | Luma AI | 42.1±8.2 | $0.040 | est.~14s | 1/6 benchmarks measured | Verified | ||||
| 21 | Luma AI | 41.0±3.2 | $0.10 | est.~29s | 4/6 benchmarks measured | Verified | ||||
| 22 | Ideogram | 36.9±4.3 | $0.10 | est.~29s | 2/6 benchmarks measured | Verified | ||||
| 23 | ByteDance Seed | 34.7±3.2 | $0.035 | est.~15s | 4/6 benchmarks measured | Verified | ||||
| 24 | ByteDance Seed | 34.6±3.2 | $0.040 | 11s | 4/6 benchmarks measured | Verified | ||||
| 25 | Black Forest Labs | 30.3±3.2 | $0.070 | 30s | 4/6 benchmarks measured | Verified | ||||
| 26 | Krea | 30.0±4.3 | $0.060 | 34s | 2/6 benchmarks measured | Verified | ||||
| 27 | Google DeepMind | 29.6±4.7 | $0.060 | est.~11s | 2/6 benchmarks measured | Verified | ||||
| 28 | Black Forest Labs | 29.6±3.2 | $0.060 | est.~23s | 4/6 benchmarks measured | Verified | ||||
| 29 | inclusionAI (Ant Group) | 29.6±5.6 | $0.030 | est.~11s | 1/6 benchmarks measured | Verified | ||||
| 30 | Black Forest Labs | 28.5±3.2 | $0.030 | 12s | 4/6 benchmarks measured | Verified | ||||
| 31 | Recraft | 28.0±5.6 | $0.035 | est.~13s | 1/6 benchmarks measured | Verified | ||||
| 32 | Recraft | 27.0±4.7 | $0.21 | est.~53s | 2/6 benchmarks measured | Verified | ||||
| 33 | FLUX.2 [dev]Open | Black Forest Labs | 26.5±3.2 | $0.012 | 16s | 4/6 benchmarks measured | Verified | |||
| 34 | HiDream-O1-ImageOpen | HiDream.ai | 24.1±3.3 | $0.040 | 7.3s | 5/6 benchmarks measured | Mixed | |||
| 35 | Black Forest Labs | 20.0±3.2 | $0.015 | 4.0s | 4/6 benchmarks measured | Verified | ||||
| 36 | Alibaba Qwen | 17.7±3.9 | $0.005 | 1.9s | 4/6 benchmarks measured | Mixed | ||||
| 37 | Qwen-ImageOpen | Alibaba Qwen | 14.2±4.0 | $0.020 | 21s | 4/6 benchmarks measured | Mixed | |||
| 38 | OpenAI | 10.5±3.2 | $0.011 | est.~6.3s | 4/6 benchmarks measured | Verified |
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
- AA Text-to-Image Arena(third-party)
- AA Image Editing Arena(third-party)
- LMArena Text-to-Image(third-party)
- LMArena Image Edit(third-party)
- GenEval(day 0)
- DPG-Bench(day 0)
Day-0 (model cards) Third-party
- Fully verified
- 92%
- of 38 models