Image Generation on ImageNet 512x512

Metric: FID (lower is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	FID▲	Extra Data	Paper	Date↕	Code
1	EDM2-L + DDO (SD-VAE, 25 steps, DPM-Solver-v3)	1.21	No	Direct Discriminative Optimization: Your Likelih...	2025-03-03	Code
2	DDT-XL/2 + UCGM-S (SD-VAE + 150 sampling steps + CFG)	1.24	No	Unified Continuous Generative Models	2025-05-12	Code
3	DDT-XL/2 + UCGM-S (SD-VAE + 100 sampling steps + CFG)	1.25	No	Unified Continuous Generative Models	2025-05-12	Code
4	EDM2-XXL Autoguidance	1.25	No	Guiding a Diffusion Model with a Bad Version of ...	2024-06-04	Code
5	DDT-XL/2(22en6de 675M + guidance interval )	1.28	No	DDT: Decoupled Diffusion Transformer	2025-04-08	Code
6	EDM2- S Autoguidance (XS, T /16)	1.34	No	Guiding a Diffusion Model with a Bad Version of ...	2024-06-04	Code
7	SiDA-EDM2-XXL (1.5B)	1.366	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
8	SiDA-EDM2-XL (1.1B)	1.379	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
9	EDM2-XXL w/ guidance interval	1.4	No	Applying Guidance in a Limited Interval Improves...	2024-04-11	Code
10	SiDA-EDM2-L (777M)	1.413	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
11	SiD2	1.48	No	Simpler Diffusion (SiD2): 1.5 FID on ImageNet512...	2024-10-25	-
12	SiDA-EDM2-M (498M)	1.488	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
13	SiDA-EDM2-S (280M)	1.669	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
14	EDM2-S w/ guidance interval	1.68	No	Applying Guidance in a Limited Interval Improves...	2024-04-11	Code
15	GMem	1.71	No	Generative Modeling with Explicit Memory	2024-12-11	Code
16	DC-AE-f32 + USiT-2B	1.72	No	Deep Compression Autoencoder for Efficient High-...	2024-10-14	Code
17	MAR-L, Diff Loss	1.73	No	Autoregressive Image Generation without Vector Q...	2024-06-17	Code
18	SIMS	1.73	No	Self-Improving Diffusion Models with Synthetic D...	2024-08-29	-
19	PaGoDA	1.8	No	PaGoDA: Progressive Growing of a One-Step Genera...	2024-05-23	Code
20	EDM2-XXL	1.81	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
21	EDM2-XL	1.85	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
22	EDM2-L	1.88	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
23	SiD-EDM2-XL (1.1B)	1.888	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
24	SiD-EDM2-L (777M)	1.907	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
25	MAGVIT-v2	1.91	No	Language Model Beats Diffusion -- Tokenizer is K...	2023-10-09	Code
26	SiD-EDM2-XXL (1.5B)	1.969	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
27	EDM2-M	2.01	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
28	SiD-EDM2-M (498M)	2.06	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
29	TiTok-B-128	2.13	No	An Image is Worth 32 Tokens for Reconstruction a...	2024-06-11	Code
30	SiDA-EDM2-XS (125M)	2.156	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
31	EDM2-S	2.23	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
32	DiT-XL/2 with CADS	2.31	No	CADS: Unleashing the Diversity of Diffusion Mode...	2023-10-26	-
33	StyleGAN-XL	2.4	No	StyleGAN-XL: Scaling StyleGAN to Large Diverse D...	2022-02-01	Code
34	TiTok-L-64	2.49	No	An Image is Worth 32 Tokens for Reconstruction a...	2024-06-11	Code
35	DiffiT	2.67	No	DiffiT: Diffusion Vision Transformers for Image ...	2023-12-04	Code
36	SiD-EDM2-S (280M)	2.707	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
37	DiT-XL/2 with SA-Solver	2.8	No	SA-Solver: Stochastic Adams Solver for Fast Samp...	2023-09-10	Code
38	DiMR-XL/3R	2.89	No	Alleviating Distortion in Image Generation via M...	2024-06-13	Code
39	EDM2-XS	2.91	No	Analyzing and Improving the Training Dynamics of...	2023-12-05	Code
40	GIVT-Causal-L+A	2.92	No	GIVT: Generative Infinite-Vocabulary Transformers	2023-12-04	Code
41	DiT-XL/2	3.04	No	Scalable Diffusion Models with Transformers	2022-12-19	Code
42	MAGVIT-v2 (w/o guidance)	3.07	No	Language Model Beats Diffusion -- Tokenizer is K...	2023-10-09	Code
43	SiD-EDM2-XS (125M)	3.353	No	Adversarial Score identity Distillation: Rapidly...	2024-10-19	Code
44	DPC-U	3.54	No	-	-	-
45	Latent Diffusion (LDM-4-G)	3.6	No	High-Resolution Image Synthesis with Latent Diff...	2021-12-20	Code
46	Poly-INR	3.81	No	Polynomial Implicit Neural Representations For L...	2023-03-20	Code
47	ADM-G, ADM-U	3.85	No	Diffusion Models Beat GANs on Image Synthesis	2021-05-11	Code
48	simple diffusion (U-Net)	4.28	No	Simple diffusion: End-to-end diffusion for high ...	2023-01-26	Code
49	MaskGIT (a=0.05)	4.46	No	MaskGIT: Masked Generative Image Transformer	2022-02-08	Code
50	simple diffusion (U-ViT, L)	4.53	No	Simple diffusion: End-to-end diffusion for high ...	2023-01-26	Code
51	MaskGIT	7.32	No	MaskGIT: Masked Generative Image Transformer	2022-02-08	Code
52	ADM-G	7.72	No	Diffusion Models Beat GANs on Image Synthesis	2021-05-11	Code

#1EDM2-L + DDO (SD-VAE, 25 steps, DPM-Solver-v3)SOTA
1.21
FID· 2025-03-03
Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator Code
#2DDT-XL/2 + UCGM-S (SD-VAE + 150 sampling steps + CFG)
1.24
FID· 2025-05-12
Unified Continuous Generative Models Code
#3DDT-XL/2 + UCGM-S (SD-VAE + 100 sampling steps + CFG)
1.25
FID· 2025-05-12
Unified Continuous Generative Models Code
#4EDM2-XXL AutoguidanceSOTA
1.25
FID· 2024-06-04
Guiding a Diffusion Model with a Bad Version of Itself Code
#5DDT-XL/2(22en6de 675M + guidance interval )
1.28
FID· 2025-04-08
DDT: Decoupled Diffusion Transformer Code
#6EDM2- S Autoguidance (XS, T /16)SOTA
1.34
FID· 2024-06-04
Guiding a Diffusion Model with a Bad Version of Itself Code
#7SiDA-EDM2-XXL (1.5B)
1.366
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#8SiDA-EDM2-XL (1.1B)
1.379
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#9EDM2-XXL w/ guidance intervalSOTA
1.4
FID· 2024-04-11
Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models Code
#10SiDA-EDM2-L (777M)
1.413
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#11SiD2
1.48
FID· 2024-10-25
Simpler Diffusion (SiD2): 1.5 FID on ImageNet512 with pixel-space diffusion
#12SiDA-EDM2-M (498M)
1.488
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#13SiDA-EDM2-S (280M)
1.669
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#14EDM2-S w/ guidance intervalSOTA
1.68
FID· 2024-04-11
Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models Code
#15GMem
1.71
FID· 2024-12-11
Generative Modeling with Explicit Memory Code
#16DC-AE-f32 + USiT-2B
1.72
FID· 2024-10-14
Deep Compression Autoencoder for Efficient High-Resolution Diffusion Models Code
#17MAR-L, Diff Loss
1.73
FID· 2024-06-17
Autoregressive Image Generation without Vector Quantization Code
#18SIMS
1.73
FID· 2024-08-29
Self-Improving Diffusion Models with Synthetic Data
#19PaGoDA
1.8
FID· 2024-05-23
PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher Code
#20EDM2-XXLSOTA
1.81
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#21EDM2-XLSOTA
1.85
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#22EDM2-LSOTA
1.88
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#23SiD-EDM2-XL (1.1B)
1.888
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#24SiD-EDM2-L (777M)
1.907
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#25MAGVIT-v2SOTA
1.91
FID· 2023-10-09
Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation Code
#26SiD-EDM2-XXL (1.5B)
1.969
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#27EDM2-M
2.01
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#28SiD-EDM2-M (498M)
2.06
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#29TiTok-B-128
2.13
FID· 2024-06-11
An Image is Worth 32 Tokens for Reconstruction and Generation Code
#30SiDA-EDM2-XS (125M)
2.156
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#31EDM2-S
2.23
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#32DiT-XL/2 with CADS
2.31
FID· 2023-10-26
CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling
#33StyleGAN-XLSOTA
2.4
FID· 2022-02-01
StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets Code
#34TiTok-L-64
2.49
FID· 2024-06-11
An Image is Worth 32 Tokens for Reconstruction and Generation Code
#35DiffiT
2.67
FID· 2023-12-04
DiffiT: Diffusion Vision Transformers for Image Generation Code
#36SiD-EDM2-S (280M)
2.707
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#37DiT-XL/2 with SA-Solver
2.8
FID· 2023-09-10
SA-Solver: Stochastic Adams Solver for Fast Sampling of Diffusion Models Code
#38DiMR-XL/3R
2.89
FID· 2024-06-13
Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models and Time-Dependent Layer Normalization Code
#39EDM2-XS
2.91
FID· 2023-12-05
Analyzing and Improving the Training Dynamics of Diffusion Models Code
#40GIVT-Causal-L+A
2.92
FID· 2023-12-04
GIVT: Generative Infinite-Vocabulary Transformers Code
#41DiT-XL/2
3.04
FID· 2022-12-19
Scalable Diffusion Models with Transformers Code
#42MAGVIT-v2 (w/o guidance)
3.07
FID· 2023-10-09
Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation Code
#43SiD-EDM2-XS (125M)
3.353
FID· 2024-10-19
Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step Code
#44DPC-U
3.54
FID
No paper
#45Latent Diffusion (LDM-4-G)SOTA
3.6
FID· 2021-12-20
High-Resolution Image Synthesis with Latent Diffusion Models Code
#46Poly-INR
3.81
FID· 2023-03-20
Polynomial Implicit Neural Representations For Large Diverse Datasets Code
#47ADM-G, ADM-USOTA
3.85
FID· 2021-05-11
Diffusion Models Beat GANs on Image Synthesis Code
#48simple diffusion (U-Net)
4.28
FID· 2023-01-26
Simple diffusion: End-to-end diffusion for high resolution images Code
#49MaskGIT (a=0.05)
4.46
FID· 2022-02-08
MaskGIT: Masked Generative Image Transformer Code
#50simple diffusion (U-ViT, L)
4.53
FID· 2023-01-26
Simple diffusion: End-to-end diffusion for high resolution images Code
#51MaskGIT
7.32
FID· 2022-02-08
MaskGIT: Masked Generative Image Transformer Code
#52ADM-GSOTA
7.72
FID· 2021-05-11
Diffusion Models Beat GANs on Image Synthesis Code