AI models and papers trending on October 21, 2026
The edition of this day as it was recorded: models with the likes and downloads they had then, and the most upvoted papers of the daily list.
On Wednesday, October 21, 2026 the model at the top of the ranking was deepseek-ai/DeepSeek-V4.1-Flash-1017 (image text to text), with 27,737 downloads in 30 days and 2,022 likes. The most upvoted paper was "Focused Omni-Modal GUI Agents" from Nanyang Technological University, with 97 votes. 11 of the 15 leading papers had a code repository attached.
Recorded 2026-10-21 07:06 UTC · numbers are those of that moment
Papers
- 97
Focused Omni-Modal GUI Agents
Nanyang Technological UniversityarXiv 2610.26203 (opens in a new tab)Code (opens in a new tab)
- 54
Sage-A1.5: From Long-Horizon Execution to Scientific Exploration with Capability Composition
Intern SciencearXiv 2610.28578 (opens in a new tab)Code (opens in a new tab)
- 53
PAR: Progress-Aligned Recurrence for Length Generalization in State Tracking
ant-internationalarXiv 2610.27628 (opens in a new tab)
- 36
Learning the 4D World through Point Trajectory Forecasting
arXiv 2610.28594 (opens in a new tab)Code (opens in a new tab)
- 33
VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
ZJU-OmniAIarXiv 2610.25043 (opens in a new tab)Code (opens in a new tab)
- 33
SolarRenderer: Few-Step Generative Reference Rendering
Show LabarXiv 2610.28630 (opens in a new tab)Code (opens in a new tab)
- 26
Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience
RopediaarXiv 2610.26855 (opens in a new tab)Code (opens in a new tab)
- 22
CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?
- 22
Correction Training for Diffusion Models
University of California, BerkeleyarXiv 2610.14136 (opens in a new tab)Code (opens in a new tab)
- 22
SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation
- 18
S-Agent 2.0: Spatial Reasoning with a General Coding Agent
RopediaarXiv 2610.28354 (opens in a new tab)Code (opens in a new tab)
- 16
MMEnterprise-Bench: Evaluating Enterprise Agents over Evolving Multimodal Records
arXiv 2610.27911 (opens in a new tab)Code (opens in a new tab)
- 15
PhysSkin++: Real-Time and Generalizable Physics-Based Animation via Self-Supervised Neural Skinning
zju3dvarXiv 2610.26067 (opens in a new tab)Code (opens in a new tab)
- 12
Breaking Diffusion: Parallel Reasoning with Draft Refinement
- 10
From Short Clips to Real-Time Streams: Identity-Aware Supervision and Context-Efficient Generation for Video Face Swapping
arXiv 2610.26681 (opens in a new tab)Code (opens in a new tab)
Models
- 1
deepseek-ai/DeepSeek-V4.1-Flash-1017
Image text to text · 27.7K downloads / 30 d · 2,022 likes
- 2
black-forest-labs/flux-3-image-klein-comfy
Image to image · 44K downloads / 30 d · 1,111 likes
- 3
Text to speech · 13.5K downloads / 30 d · 848 likes
- 4
meta-models/Muse-Spark-Open-816B-A42B
Image text to text · 2,592 downloads / 30 d · 613 likes
- 5
black-forest-labs/flux-3-image-klein
Image to image · 78.9K downloads / 30 d · 574 likes
- 6
Feature extraction · 172K downloads / 30 d · 2,777 likes
- 7
Text to image · 4,547 downloads / 30 d · 506 likes
- 8
Image to video · 2.04M downloads / 30 d · 8,653 likes
- 9
Text generation · 115K downloads / 30 d · 2,032 likes
- 10
Text to image · 3,558 downloads / 30 d · 440 likes
- 11
abenzerps/Qwen-Image-2.1-Uncensored-Quantized
Text to image · 2.79M downloads / 30 d · 4,729 likes
- 12
LiquidAI/LFM2.5-24B-A2B (opens in a new tab)
Text generation · 15.4K downloads / 30 d · 530 likes
- 13
Image text to text · 7.59M downloads / 30 d · 18.1K likes
- 14
No task tag · 925K downloads / 30 d · 306 likes
- 15
nvidia/Qwen3.8-Flash-Next-FP4-Local
Image text to text · 7,345 downloads / 30 d · 274 likes
About the archive
How far back does the archive go?
Model lists exist from the first day the collector ran: Hugging Face does not publish past trending lists, so they cannot be rebuilt. Paper lists for earlier days were rebuilt from the Hugging Face daily papers archive, and those days are marked "papers only".
What is kept for each day?
The leading models with their task, likes and downloads on that day, the models held back, and the leading papers with their upvotes. Abstracts are not duplicated in the archive; each paper links to arXiv.
Do archived numbers change?
No. A day’s record is the last collection of that day and is not rewritten afterwards. Likes and downloads you see on an archived day are the values on that day, not today’s.
How are weeks defined?
ISO 8601 weeks: Monday to Sunday, numbered so that week 1 contains the first Thursday of the year. The address /week/2026-40 is week 40 of 2026.