ImageBench
Text rendering benchmark

Best AI image generator for text rendering

Text in generated images is still a practical failure point: spelling, small labels, poster copy, typography, and layout all matter. ImageBench V1 ranks models on the same text-rendering prompts so the comparison is evidence-based.

Current leader: openai/gpt-image-2.5-sunburst at 100.0%.

openai/gpt-image-2 text rendering benchmark example
openai/gpt-image-2
100.0% text score
fal/google/nano-banana-2 text rendering benchmark example
fal/google/nano-banana-2
73.3% text score
bfl/flux-2-max text rendering benchmark example
bfl/flux-2-max
93.3% text score
bfl/flux-2-pro text rendering benchmark example
bfl/flux-2-pro
80.0% text score

Text rendering ranking

This score covers typography style, spelling accuracy, multi-line copy, posters, product labels, and small readable text.

RankModelText renderingOverallPass / Fail
1openai/gpt-image-2.5-sunburst100.0%88.0%169 / 23
2openai/gpt-image-2.5-flare100.0%86.5%166 / 26
3openai/gpt-image-2100.0%87.0%167 / 25
4xai/grok-imagine-2100.0%79.2%152 / 40
5fal/reve/2.1100.0%79.7%153 / 39
6fal/google/nano-banana-pro100.0%78.6%151 / 41
7fal/qwen-image-2-pro100.0%75.5%145 / 47
8fal/luma/uni-1-max100.0%78.6%151 / 41
9fal/bytedance/seedream-v5-pro93.3%75.0%144 / 48
10bfl/flux-2-max93.3%67.2%129 / 63
11fal/bytedance/seedream-v5-lite93.3%76.6%147 / 45
12local/qwen-image-2.1-7b-nf493.3%64.1%123 / 69
13local/sefi-image-5b-rl93.3%60.9%117 / 75
14fal/ideogram/v493.3%62.0%119 / 73
15fal/bytedance/seedream-v486.7%68.2%131 / 61
16local/moss-image-sft86.7%66.1%127 / 65
17local/flux-3-klein-fp886.7%60.4%116 / 76
18local/flux-3-klein-bf1686.7%59.9%115 / 77
19replicate/recraft-ai/recraft-v486.7%62.5%120 / 72
20local/sefi-image-5b-base86.7%62.5%120 / 72
21local/moss-image-base-dmd286.7%60.4%116 / 76
22fal/microsoft/mai-image-2.580.0%69.3%133 / 59
23local/sensenova-u1.5-8b-mot-base80.0%70.3%135 / 57
24local/boogu-image-turbo80.0%62.0%119 / 73
25bfl/flux-2-pro80.0%67.2%129 / 63
26local/flux-3-klein-nvfp480.0%64.6%124 / 68
27local/flux-3-klein-base-nvfp480.0%64.1%123 / 69
28local/flux-3-klein-base-fp880.0%62.5%120 / 72
29local/flux-3-klein-base-bf1680.0%60.4%116 / 76
30fal/krea/v2-large80.0%64.1%123 / 69
31local/moss-image-base80.0%57.3%110 / 82
32fal/google/nano-banana-273.3%77.1%148 / 44
33replicate/google/imagen-4-ultra73.3%62.0%119 / 73
34fal/hidream-o1-image-8b73.3%52.6%101 / 91
35local/minimax-h3-image-33b73.3%62.0%119 / 73
36fal/krea/v2-medium-turbo73.3%60.9%117 / 75
37local/kromoss73.3%61.5%118 / 74
38local/mage-flow-rl73.3%51.0%98 / 94
39local/krea-2-raw73.3%61.5%118 / 74
40local/ming-image-0.1-design-6b66.7%61.5%118 / 74
41local/sensenova-u1.5-8b-mot-8step66.7%63.5%122 / 70
42local/kromoss-dmd266.7%64.6%124 / 68
43local/kromoss-dmd66.7%62.5%120 / 72
44fal/ideogram/v366.7%48.4%93 / 99
45fal/hidream-o1-image-dev-8b66.7%46.4%89 / 103
46local/mage-flow-turbo66.7%54.2%104 / 88
47local/z-image-6b66.7%56.8%109 / 83
48fal/krea/v2-medium66.7%62.5%120 / 72
49bfl/flux-1.1-pro-ultra66.7%45.3%87 / 105
50fal/recraft/v366.7%41.1%79 / 113
51local/krea-2-turbo60.0%63.0%121 / 71
52local/krea-2-turbo-no-filter60.0%58.3%112 / 80
53local/iris-3b60.0%53.1%102 / 90
54local/z-image-turbo-6b60.0%52.6%101 / 91
55local/mage-flow-base60.0%50.0%96 / 96
56local/qwen-image-2512-20b53.3%52.1%100 / 92
57fal/hunyuan-image-353.3%52.1%100 / 92
58bfl/flux-1-dev53.3%45.8%88 / 104
59local/iris-3b-turbo53.3%47.4%91 / 101
60replicate/leonardoai/lucid-origin53.3%52.1%100 / 92
61bfl/flux-1.1-pro53.3%45.8%88 / 104
62local/sefi-image-5b-turbo53.3%56.8%109 / 83
63replicate/luma/photon53.3%51.6%99 / 93
64local/nucleus-image-17b-a2b53.3%44.3%85 / 107
65local/flux-2-klein-9b46.7%57.8%111 / 81
66replicate/google/imagen-4-fast46.7%51.6%99 / 93
67replicate/minimax/image-0146.7%50.0%96 / 96
68local/prxpixel-t2i-7b46.7%46.4%89 / 103
69local/hidream-i1-full-17b46.7%40.1%77 / 115
70replicate/luma/photon-flash46.7%45.8%88 / 104
71bfl/flux-2-klein-9b40.0%57.3%110 / 82
72local/flux-2-klein-4b40.0%52.6%101 / 91
73bfl/flux-2-klein-4b40.0%50.5%97 / 95
74local/mgflow-flux2-klein-4b33.3%43.2%83 / 109
75fal/bria/fast33.3%56.2%108 / 84
76fal/flux-1-schnell33.3%37.0%71 / 121
77local/bonsai-image-ternary-4b26.7%45.3%87 / 105
78replicate/google/imagen-420.0%46.4%89 / 103
79local/sefi-image-2b-turbo20.0%44.3%85 / 107
80local/sana-1.5-1.6b13.3%34.9%67 / 125
81local/tinydit-2560.0%21.9%42 / 150
82local/supra2-img-100m0.0%14.1%27 / 165
83local/tld-101m0.0%13.0%25 / 167

Examples from the benchmark

openai/gpt-image-2

100.0% text score, 15/15 pass

Details
openai/gpt-image-2 text rendering benchmark example

A poster with headline 'SUMMER SALE' in large bold text, body text 'Up to 50% off all items this weekend only' in medium size, and fine print 'Terms and conditions apply. See store for details.' at the bottom in small text

PASS

fal/google/nano-banana-2

73.3% text score, 11/15 pass

Details
fal/google/nano-banana-2 text rendering benchmark example

A poster with headline 'SUMMER SALE' in large bold text, body text 'Up to 50% off all items this weekend only' in medium size, and fine print 'Terms and conditions apply. See store for details.' at the bottom in small text

PASS

bfl/flux-2-max

93.3% text score, 14/15 pass

Details
bfl/flux-2-max text rendering benchmark example

A poster with headline 'SUMMER SALE' in large bold text, body text 'Up to 50% off all items this weekend only' in medium size, and fine print 'Terms and conditions apply. See store for details.' at the bottom in small text

PASS

bfl/flux-2-pro

80.0% text score, 12/15 pass

Details
bfl/flux-2-pro text rendering benchmark example

A poster with headline 'SUMMER SALE' in large bold text, body text 'Up to 50% off all items this weekend only' in medium size, and fine print 'Terms and conditions apply. See store for details.' at the bottom in small text

PASS

When text rendering matters

Product packaging and labels
Posters, flyers, and ads
Charts, diagrams, and UI mockups
Book covers and editorial graphics
See the full benchmark