AI Paper Atlas

Research map / Benchmarks / ImageNet 256x256 class-conditional

ImageNet 256x256 class-conditional leaderboard

37 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ImageNet 256x256 class-conditional. Best FID: 1.22 (XYZFlow-H (w/ GAN), ICML 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

ImageNet 256x256 class-conditional · FID

33 papers, each with its best reported number for its own method; lower is better.

#FID ↓MethodPaperTopic
11.22XYZFlow-H (w/ GAN)XYZFlow: Scaling Multidimensional Shortcut Flows for Efficient Generative Modeling
ICML 2026 · Table 1
Fast diffusion sampling
21.24BitDance-H-1x
BitDance-H (1.0B)
BitDance: Scaling Autoregressive Generative Models with Binary Tokens
NeurIPS 2026 · Table 2
Autoregressive and latent generation
31.27AliTok-XL + SoftCFG (tuned gamma, k)
AliTok-XL (662M)
SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model
ICLR 2026 · Table 1
Autoregressive and latent generation
41.31Ours (MAR)
MAR (943M)
Condition Errors Refinement in Autoregressive Image Generation with Diffusion Loss
ICLR 2026 · Table 1
Fast diffusion sampling
51.34SphereAR-H (943M)Hyperspherical Latents Improve Continuous-Token Autoregressive Generation
ICLR 2026 · Table 1
Autoregressive and latent generation
61.57RAR-L + MASC (4k)
RAR-L + GigaTok tokenizer
A Flat Vocabulary or a Rich Hierarchy? Re-introducing Intrinsic Structure Transforms the Autoregressive Image Generation
ICML 2026 · Table 4
Autoregressive and latent generation
71.62CrossFlow-XL (1 NFE)Cross Flow: One-Step Generation Across Latent and Pixel Spaces
NeurIPS 2026 · Table 1
Fast diffusion sampling
81.69LazyMAR-H, 64 stepsLazyMAR: Accelerating Masked Autoregressive Models via Feature Caching
ICCV 2025 · Table 1
Autoregressive and latent generation
91.79Composer
VAR d-30
Test-Time Instance-Specific Parameter Composition: A New Paradigm for Adaptive Generative Modeling
CVPR 2026 · Table 2
Reasoning and chain of thought
101.81GRN-GGenerative Refinement Network for Visual Synthesis
ECCV 2026 · Table 3
Autoregressive and latent generation
23 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ImageNet 256x256 class-conditional · IS

16 papers, each with its best reported number for its own method; higher is better.

#IS ↑MethodPaperTopic
1349.2ARPG-XL (32 steps)
ARPG-XL
Autoregressive Image Generation with Randomized Parallel Decoding
ICLR 2026 · Table 1
Autoregressive and latent generation
2338.7SCAN-XXL (30 refine)SCAN: Self-Calibrated AutoregressioN for High-Quality Visual Generation
AAAI 2026 · Table 1
Autoregressive and latent generation
3330.5VAR-d30 + AMS-KV
VAR-d30
AMS-KV: Adaptive KV Caching in Multi-Scale Visual Autoregressive Transformers
AAAI 2026 · Table 3
Attention and transformers
4324.2Ours (MAR)
MAR (943M)
Condition Errors Refinement in Autoregressive Image Generation with Diffusion Loss
ICLR 2026 · Table 1
Fast diffusion sampling
5319.5Composer
VAR d-24
Test-Time Instance-Specific Parameter Composition: A New Paradigm for Adaptive Generative Modeling
CVPR 2026 · Table 2
Reasoning and chain of thought
6318.5ReDDiT-XL (f8)ReDDiT: Rehashing Noise for Discrete Visual Generation
ICLR 2026 · Table 1
Fast diffusion sampling
7317NVG-d24Next Visual Granularity Generation
ICLR 2026 · Table 1
Autoregressive and latent generation
8304.4BitDance-H-1x
BitDance-H (1.0B)
BitDance: Scaling Autoregressive Generative Models with Binary Tokens
NeurIPS 2026 · Table 2
Autoregressive and latent generation
9302.4AliTok-XL + SoftCFG (tuned gamma, k)
AliTok-XL (662M)
SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model
ICLR 2026 · Table 1
Autoregressive and latent generation
10300SphereAR-H (943M)Hyperspherical Latents Improve Continuous-Token Autoregressive Generation
ICLR 2026 · Table 1
Autoregressive and latent generation
6 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.