AI models and papers of week 43, 2026
6 recorded days folded into one page: the 20 most upvoted papers of the week and the models that held a place on the trending list, from 6 days of model records.
The paper with the most votes this week was “WeVisOne-VL: Unified Visual Perception from Objects to Pixels” from Tencent, with 226 upvotes on the list of October 23, 2026. 60 of the 81 papers recorded came with a code repository.
Papers of the week
- 226
- 165
- 152
- 97
Focused Omni-Modal GUI Agents
Wed Oct 21 · Nanyang Technological UniversityarXiv 2610.26203 (opens in a new tab)Code (opens in a new tab)
- 88
Has GPT-6-Astra Achieved Embodied Intelligence?
Mon Oct 19 · SenseNovaarXiv 2610.22471 (opens in a new tab)Code (opens in a new tab)
- 86
Pack Light, CarryOn: Building Object-Oriented Context for Long-Horizon Agents
Mon Oct 19 · McAuley-LabarXiv 2610.18784 (opens in a new tab)Code (opens in a new tab)
- 84
Learning from Fleet Experience: A Collective Learning Paradigm for Robots
Tue Oct 20arXiv 2610.25736 (opens in a new tab)
- 74
- 69
HexaShell: A Thin Harness for Exploring the Limits of VLM Agents as Robot Coding Agents
Mon Oct 19 · HexaFuturearXiv 2610.23479 (opens in a new tab)Code (opens in a new tab)
- 69
Into the Unknown: Benchmarking AI agents on open scientific research problems
Mon Oct 19 · Carnegie Mellon UniversityarXiv 2610.19159 (opens in a new tab)Code (opens in a new tab)
- 57
OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory
Tue Oct 20 · University of California, Los AngelesarXiv 2610.26860 (opens in a new tab)Code (opens in a new tab)
- 56
- 54
BizBench: Unveiling Hidden Challenges in Deploying Long-Horizon Agents for Real-World Business
Mon Oct 19 · ant-internationalarXiv 2610.20424 (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
Wed Oct 21 · Intern SciencearXiv 2610.28578 (opens in a new tab)Code (opens in a new tab)
- 54
Cache Engineering for Agents
Thu Oct 22 · Tsinghua-NICS-EFCarXiv 2610.28577 (opens in a new tab)Code (opens in a new tab)
- 53
PAR: Progress-Aligned Recurrence for Length Generalization in State Tracking
Wed Oct 21 · ant-internationalarXiv 2610.27628 (opens in a new tab)
- 53
Agent World Model in the Wild: Computer-Use Agent Environment Synthesis
Thu Oct 22 · Microsoft ResearcharXiv 2610.26828 (opens in a new tab)Code (opens in a new tab)
- 51
Looking Ahead: Modeling Future Evolution in Autoregressive Video Generation
Thu Oct 22 · ARC Lab, TencentarXiv 2610.29989 (opens in a new tab)Code (opens in a new tab)
- 49
Toward Recursive Self-Improvement in Physical AI
Fri Oct 23 · National University of SingaporearXiv 2610.29677 (opens in a new tab)Code (opens in a new tab)
- 40
ScaleNav: Scalable Internet Video Pretraining for Language-Conditioned Local Navigation
Tue Oct 20arXiv 2610.25690 (opens in a new tab)Code (opens in a new tab)
Models of the week
deepseek-ai/DeepSeek-V4.1-Flash-1017
6 of 6 days · best No. 1 · Image text to text · 57.7K downloads / 30 d
black-forest-labs/flux-3-image-klein-comfy
6 of 6 days · best No. 2 · Image to image · 535K downloads / 30 d
6 of 6 days · best No. 3 · Feature extraction · 193K downloads / 30 d
6 of 6 days · best No. 3 · Text to speech · 17K downloads / 30 d
6 of 6 days · best No. 4 · Text to image · 7,658 downloads / 30 d
6 of 6 days · best No. 5 · Image to video · 2.03M downloads / 30 d
6 of 6 days · best No. 5 · Text to image · 5,618 downloads / 30 d
abenzerps/Qwen-Image-2.1-Uncensored-Quantized
6 of 6 days · best No. 9 · Text to image · 2.4M downloads / 30 d
meta-models/Muse-Spark-Open-816B-A42B
5 of 6 days · best No. 3 · Image text to text · 3,163 downloads / 30 d
5 of 6 days · best No. 5 · Text generation · 125K downloads / 30 d
5 of 6 days · best No. 10 · Image text to text · 7.71M downloads / 30 d
black-forest-labs/flux-3-image-klein
4 of 6 days · best No. 4 · Image to image · 92.6K downloads / 30 d
4 of 6 days · best No. 8 · No task tag · 1.61M downloads / 30 d
3 of 6 days · best No. 2 · Text generation · 1,904 downloads / 30 d
LiquidAI/LFM2.5-24B-A2B (opens in a new tab)
3 of 6 days · best No. 8 · Text generation · 15.4K downloads / 30 d
openbmb/MiniCPM-V-4.7-35B-A3B (opens in a new tab)
2 of 6 days · best No. 2 · Image text to text · 76.4K downloads / 30 d
2 of 6 days · best No. 6 · Text generation · 1,060 downloads / 30 d
baidu/Unlimited-OCR (opens in a new tab)
2 of 6 days · best No. 13 · Image text to text · 1.8M downloads / 30 d
nvidia/Qwen3.8-Flash-Next-FP4-Local
2 of 6 days · best No. 15 · Image text to text · 9,340 downloads / 30 d
dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8
1 of 6 days · best No. 13 · Image text to text · 4,141 downloads / 30 d
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.