whatmodel.sh / image / GPT Image 1.5

GPT Image 1.5

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

GPT Image 1.5 is a closed-weights image model from OpenAI, released December 2025. Licence: proprietary.

71.2±1.4#15 of 144 image modelsfrom 12 benchmark results across 11 sources

Score within the LMArena text-to-image comparison set only.

$ how it compares

The image models ranked either side of GPT Image 1.5, out of 197 with a score, then more from OpenAI.

#modelscorewhereRAMreleased
15MAI-Image-2.5 Microsoft AI74±2comparison only2026-06
16Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image) Google73±2api $0.034/image2026-06
17GPT Image 1.5 OpenAI71±1api $0.034/image2025-12
18Qwen-Image-2.1 Alibaba69±2local~12GB2026-09
19Ideogram 4.5 Ideogram68±3comparison only
more from OpenAI
1GPT Image 2.5 Sunburst OpenAI92±2api $0.013/image2026-09
2GPT Image 2.5 Flare OpenAI88±2api $0.013/image2026-09
3GPT Image 2 OpenAI85±1api $0.053/image2026-04

see all image models · find the ones that fit your machine

$ what it's good at

text to image▂▅▅▃▄▄█▃▃▂▁▁ #14 of 126 · 68% of its score · 5 tests
testresultvs the others in its rundifficultycountswhy
GenExam (relaxed overall, GPT-5 judge)81.5% ✎▄█▂▂▂▂▆▄·▃·▂ beats 15/16hard · avg 26%▇▇▇▇▇▇▇▇▇▇▇▇ 29%sharp
LMArena text-to-image1237.4 ✎▂▂▃▄▇█▄▃▄▂·▂ beats 61/75avg 1027▇▇▇▇▇▇▇▇▇▇▇ 28%typical
UltraText Bench (Composite, bilingual dense text rendering)58.0% ✎▁▂▂█▃▁▂▃·▁·▁ beats 13/22hard · avg 28%▇▇ 4.4%sharp · noisy
FEPBench instruction faithfulness59.8% ✎▄·█···▄▄▄▄▄▄ beats 6/8medium · avg 36%▇ 3.6%blunt · at its level
Qwen-Image-Bench (Q-Judger, v1)59.7% ✎█▅▅·▅▅█·█··▃ beats 15/17medium · avg 46%▇ 2.9%blunt · at its level
image to image / editing▁▁·▃▂▃█▃▃▄▁▁ #15 of 62 · 24% of its score · 5 tests

Beats Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image) here, though they rank higher overall. Trails Grok Imagine Image Pro, Qwen-Image-2.1 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Arena image edit1369.7 ✎▂▁▁▂▆▇█▄▆▅▁▂ beats 40/53avg 1153▇▇▇▇ 11%typical
GEBench GUI state generation (English GE-Score)63.2% ✎▄▄····▄▄█▄·█ beats 8/8hard · avg 34%▇▇ 5.1%at its level
GEditBench v2 (overall Elo)1071.0 ✎▃▂····▂█▂▃▃▃ beats 12/14avg 956▇▇ 4.1%blunt
MultiTextEdit cross-lingual text-in-image editing (SemAvg, LVM judge)71.5% ✎▄·▄▄▄▄··█▄·▄ beats 7/8medium · avg 50%▇ 2.5%at its level
TECCI human overall success rate (IRCS+GGIS)5.7% ✎▄········▄█▄ beats 0/4very hard · avg 7.1%▇ 1.4%typical
other tests2 not in a task · 7.9% of its score
testresultvs the others in its rundifficultycountswhy
Rapidata Benchmark.ai image benchmark (overall Elo)1131.9 ✎▁▁▁▄▂█▂▁▁▃▁▁ beats 30/34avg 875▇▇ 4.3%typical
EdiVal multi-turn editing overall (EdiVal-O after turn 3)59.6% ✎▂·▂▃·▂·▃▂·▃█ beats 13/14medium · avg 42%▇ 3.6%blunt · at its level

$ where to run it

Through an API

api · $0.034/image · developers.openai.com ↗ · price sourceUSD for one generated 1024×1024 medium-quality image. Output-only published price; input text/image token charges are additional. Scheduled deprecation 2026-12-01. Product: Image generation API. API model: gpt-image-1.5. Settings: quality=medium, width=1024, height=1024, n=1. Price basis: generated image output component. Additional/excluded charges: input text/image tokens. Currency: USD.

$ how the score was worked out · LMArena leaderboard dataset 38%, GenExam: A Multidisciplinary… 29%, GEBench: Benchmarking Image G… 5% and 8 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
LMArena leaderboard dataset2▇▇▇▇▇▇▇▇▇▇▇ 38%CC BY 4.0
GenExam: A Multidisciplinary Text-to-Image Exam (arXiv v5)1▇▇▇▇▇▇▇▇▇ 29%CC-BY-4.0
GEBench: Benchmarking Image Generation Models as GUI Environments1▇▇ 5%CC-BY-4.0
UltraText Bench: A Comprehensive Bilingual Benchmark for Evaluating Visual Text Rendering in Image Generation (arXiv:2610.09823v1)1▇ 4%CC-BY-4.0
Rapidata/text-to-image-human-preference-6M (Benchmark.ai Image Benchmark) published dataset-card rankings1▇ 4%CC-BY-4.0
GEditBench v2: A Human-Aligned Benchmark for General Image Editing1▇ 4%CC-BY-4.0
EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing (arXiv:2509.13399v3)1▇ 4%CC-BY-4.0
Faithful, Enriched, and Precise: Benchmarking Natural-Science Illustration Generation by T2I models (FEPBench)1▇ 4%CC0-1.0
Qwen-Image-Bench: From Generation to Creation in Text-to-Image Evaluation1▇ 3%CC-BY-4.0
MultiTextEdit: Benchmarking Cross-Lingual Degradation in Text-in-Image Editing (arXiv:2605.08163v2), Table 31▇ 3%CC BY 4.0
TECCI: Tricky Edits of Collected and Curated Images1▇ 1%CC-BY-4.0
prior · pull toward the middle▇ 1%
weighted average= 71.2

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json