whatmodel.sh / image / DALL·E 3

DALL·E 3

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

DALL·E 3 is a closed-weights image model from OpenAI, released October 2023.

Model information sources

Release date ↗

44.4±1.3comparison only: no current run option verifiedfrom 13 benchmark results across 10 sources

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

$ how it compares

The image models ranked either side of DALL·E 3, out of 197 with a score, then more from OpenAI.

#modelscorewhereRAMreleased
123Show-o2 1.5B Show Lab, NUS45±5comparison only2025-06
124DeepGen 1.0 DeepGen Team45±3comparison only2026-02
125DALL·E 3 OpenAI44±1comparison only2023-10
126FLUX.1 Kontext [dev] Black Forest Labs44±1api $0.025/image +local~21GB2025-06
127FLUX.1 [dev] Black Forest Labs44±1api $0.025/image +local~7.7GB2024-08
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▁▂▇▇▆▅█▇▂▃▁▁ #96 of 157 · 99% of its score · 13 tests

Beats Latent Diffusion text2img-large (LDM-KL-8-G, LAION-400M), VQ-Diffusion (ITHQ) here, though they rank higher overall. Trails UNO here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
LMArena text-to-image967.8 ✎▂▃▃▄▇█▄▂▄▂·▂ beats 3/75avg 1027▇▇▇▇▇▇▇▇▇▇▇ 27%typical
TIIF-Bench v3 (short prompts, TIIF-Evaluator)72.2% ✎▂█····▄▄▄··▂ beats 3/11easy · avg 79%▇▇▇▇▇ 12%at its level
EvalMuse-40K model ranking (FGA-BLIP2 overall alignment)3.63 / 5 ✎█▄·▄██▄▄█▄▄▄ beats 14/14easy · avg 3.83 / 5▇▇▇▇▇ 12%at its level
ConceptMix (k=2 full-mark score)61.0% ✎▄·▄·█▄▄▄···▄ beats 7/7easy · avg 66%▇▇▇▇ 11%at its level
GenEval62.7% ✎██·▄·▄·█···▄ beats 3/8easy · avg 71%▇▇▇▇ 9.5%at its level
GeckoNum Task 1: exact number generation (human-annotated)45.2% ✎██·█·······█ beats 3/3medium · avg 51%▇▇▇ 6.4%at its level
ImagenHub text-guided generation (human overall)76.0% ✎▄▄·█·······▄ beats 4/4easy · avg 80%▇▇ 4.8%at its level
MMMG image generation (average over 15 tasks)35.2% ✎█··█▄▄·▄▄·▄▄ beats 1/9medium · avg 51%▇▇ 4.7%at its level
5 more tests
testresultvs the others in its rundifficultycountswhy
Human text alignment · wins against Kandinsky 3.050.0% ✎█·█········█ beats 2/2medium · avg 58%▇▇ 4.0%at its level · small run
Human visual quality · wins against Kandinsky 3.056.0% ✎█····█·····█ beats 2/2easy · avg 66%▇ 3.6%at its level · small run
KolorsPrompts human overall satisfaction (Kolors team)3.32 / 5 ✎▄·▄········█ beats 0/3easy · avg 3.87 / 5▇ 3.1%at its level
GenAI-Bench (VQAScore overall)81.0% ✎█···█·█····█ beats 2/3easy · avg 79%▇ 1.1%blunt
REAL realism benchmark for T2I (average of attribute, relationship and style scores)52.4% ✎█··········▄ beats 0/2easy · avg 65%▇ 0.9%blunt · at its level · small run

$ where to run it

No verified API or download recorded yet · add one

$ how the score was worked out · LMArena leaderboard dataset 27%, MMMG: a Comprehensive and Rel… 14%, TIIF-Bench: How Does Your T2I… 13% and 7 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 dataset1▇▇▇▇▇▇▇▇ 27%CC BY 4.0
MMMG: a Comprehensive and Reliable Benchmark for Multitask Multimodal Generation (arXiv:2505.17613v2)2▇▇▇▇ 14%CC-BY-4.0
TIIF-Bench: How Does Your T2I Model Follow Your Instructions? (arXiv v3)2▇▇▇▇ 13%CC-BY-4.0
EvalMuse-40K: A Reliable and Fine-Grained Benchmark with Comprehensive Human Annotations for Text-to-Image Generation Model Evaluation (arXiv v2)1▇▇▇▇ 12%CC-BY-4.0
ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficulty (arXiv v1)1▇▇▇ 11%CC-BY-4.0
Kandinsky 3.0 — original human comparisons2▇▇ 8%CC BY 4.0
Evaluating Numerical Reasoning in Text-to-Image Models (GeckoNum, arXiv v3)1▇▇ 6%CC-BY-4.0
ImagenHub: Standardizing the evaluation of conditional image generation models (arXiv v4)1▇ 5%CC-BY-4.0
Kwai-Kolors/Kolors repository README, human assessment1▇ 3%Apache-2.0
REAL: Realism Evaluation of Text-to-Image Generation Models for Effective Data Augmentation (arXiv:2502.10663v1)1▇ 1%CC-BY-4.0
prior · pull toward the middle▇ 1%
weighted average= 44.4

$ 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