Skip to content

Archive of daily editions

Archive · September 28, 2026 to October 24, 2026

Every edition keeps its date and its address. Hugging Face shows only the current trending list; here you can open what it looked like on a given day.

Recorded days run from September 28, 2026 to October 24, 2026.

By week

21 days shipped with this build

Week 43, 2026

October 19, 2026 – October 25, 2026

The week in review
  1. Saturday, October 24, 2026

    Top paper · 226 votes WeVisOne-VL: Unified Visual Perception from Objects to PixelsTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  2. Friday, October 23, 2026

    Top paper · 91 votes Deep Search in Real-World Web EnvironmentsTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  3. Thursday, October 22, 2026

    Top paper · 54 votes Cache Engineering for AgentsTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  4. Wednesday, October 21, 2026

    Top paper · 97 votes Focused Omni-Modal GUI AgentsTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  5. Tuesday, October 20, 2026

    Top paper · 84 votes Learning from Fleet Experience: A Collective Learning Paradigm for RobotsTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  6. Monday, October 19, 2026

    Top paper · 88 votes Has GPT-6-Astra Achieved Embodied Intelligence?Top model deepseek-ai/DeepSeek-V4.1-Flash-1017

Week 42, 2026

October 12, 2026 – October 18, 2026

The week in review
  1. Sunday, October 18, 2026

    Top paper · 666 votes Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy DistillationTop model deepseek-ai/DeepSeek-V4.1-Flash-1017

  2. Saturday, October 17, 2026

    Top paper · 660 votes Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy DistillationTop model google/embeddinggemma-2

  3. Friday, October 16, 2026

    Top paper · 129 votes OmniCompass: A Holistic Benchmark for Omni-Modal Video Grounding and UnderstandingTop model google/embeddinggemma-2

  4. Thursday, October 15, 2026

    Top paper · 113 votes Player-1: Learning to Play from Synthetic ExperienceTop model google/embeddinggemma-2

  5. Wednesday, October 14, 2026

    Top paper · 50 votes Urbex: Agentic Spatial Grounding for City-Scale 3D ScenesTop model google/embeddinggemma-2

  6. Tuesday, October 13, 2026

    Top paper · 65 votes Scaling Agentic Data for Long-Horizon Terminal IntelligenceTop model google/embeddinggemma-2

  7. Monday, October 12, 2026

    Top paper · 41 votes ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction AgentsTop model google/embeddinggemma-2

Week 41, 2026

October 5, 2026 – October 11, 2026

The week in review
  1. Sunday, October 11, 2026

    Top paper · 149 votes AgentGarten: Code Worlds for Evolving AgentsTop model google/embeddinggemma-2

  2. Saturday, October 10, 2026

    Top paper · 145 votes AgentGarten: Code Worlds for Evolving AgentsTop model google/embeddinggemma-2

Week 40, 2026

September 28, 2026 – October 4, 2026

The week in review
  1. Saturday, October 3, 2026

    Top paper · 152 votes OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video InteractionTop model Cloudflare/clef

  2. Friday, October 2, 2026

    Top paper · 152 votes OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video InteractionPapers only: no model list was recorded.

  3. Thursday, October 1, 2026

    Top paper · 372 votes The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy DistillationPapers only: no model list was recorded.

  4. Wednesday, September 30, 2026

    Top paper · 515 votes Raven: The Harness of Harnesses for Composable Agentic IntelligencePapers only: no model list was recorded.

  5. Tuesday, September 29, 2026

    Top paper · 271 votes Post-Training Leaves Behavioral Shadows on Unrelated DecisionsPapers only: no model list was recorded.

  6. Monday, September 28, 2026

    Top paper · 158 votes FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation AutoencodersPapers only: no model list was recorded.

Why keep an archive

A trending list is erased by the next one. Hugging Face’s API returns the list as it stands now and offers no way to ask what it was last Tuesday. The only way to have that record is to save it every day, which is what the collector does.

What the record lets you do

Check when a model first appeared and how long it stayed. See which papers led on the day a model was released. Compare the likes and downloads a repository had on two dates. Cite a stable address for “what was trending on that day” instead of a page that changes under you.

Two kinds of day

Days recorded by the collector hold both lists with the numbers of that day. Days before the collector started hold papers only: Hugging Face keeps its daily paper lists, so those were rebuilt after the fact and are marked as such. Vote counts on rebuilt days are the counts at the time of rebuilding.

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.