Skip to content

nvidia-cosmos/cosmos-reason1

View on GitHub ↗Related repositories →

Cosmos-Reason1 models understand the physical common sense and generate appropriate embodied decisions in natural language through long chain-of-thought reasoning processes.

active 2025-03-192026-06-07 (UTC)

Partial coverage11,369 / 12,256 hourly files (93%) · 2 absent upstream · 884 failed, retryable2025-03-182026-08-10 (UTC)— sampled evenly across the window, so rankings and trends hold; absolute counts scale up.
Events
683
Pushes
46
Pull requests
41
Issues
49
Stars
441
Forks
30

Activity over time

Daily event counts in the loaded window

Line chart, 446 days from 2025-03-19 to 2026-06-07. Pushes: 46 total, peak 12 in a day. Pull requests: 41 total, peak 7 in a day. Issues: 49 total, peak 6 in a day. Comments: 50 total, peak 4 in a day. Stars: 441 total, peak 21 in a day.

  • Pushes
  • Pull requests
  • Issues
  • Comments
  • Stars

Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.

Top contributors

Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity

ContributorContributionsPushesPRsComments
spectralflight4310235
bddppq231614
sophiahhuang12012
jingyijin211331
zekunhao1995101000
tylin8015
jcao-ai8122
pjannaty5012
LiuLi19985003
yangyioryy4001
doruksonmez4002
zhhezhhe4003
yfding163001
mingyuliutw3201
2jungg3000
shunzh3003
tyxiong232000
Jaykumaran2001
yifan162200
paularamo2002

Recent activity

Latest issues, pull requests and releases

  • Releasesophiahhuang2025-12-20 01:07
    Cosmos-Reason2 is out 🎉
  • Issue#852jungg2025-12-14 23:29
    Can Cosmos-Reason1 model be quantized to NVFP4 format?
  • Issue#852jungg2025-12-14 23:29
    Can Cosmos-Reason1 model be quantized to NVFP4 format?
  • Issue#852jungg2025-12-14 23:29
    Can Cosmos-Reason1 model be quantized to NVFP4 format?
  • Issue comment#83paularamo2025-12-09 21:45
    inference: add CLI overrides for fps/total_pixels and vLLM memory knobs to prevent CUDA OOM in video runs
  • Pull request#84helenzhou05232025-12-08 23:11
  • Issue comment#78WeijieH2025-12-06 16:39
    torchvision Fails to Read Video Frames (Reports 0 Frames) on linux-aarch64 (DGX Spark)
  • Issue comment#80mozhengzy2025-11-28 03:31
    Clarification on Frame Indexing and Clip Segmentation in COSMOS-REASON1
  • Issue comment#76takanotaiga2025-11-20 08:52
    Cosmos-Reason1-7B, Blackwell PRO 6000, Invalid Argument Error on vLLM 0.10.2
  • Issue comment#83paularamo2025-11-18 10:56
    inference: add CLI overrides for fps/total_pixels and vLLM memory knobs to prevent CUDA OOM in video runs
  • Issue comment#81pjannaty2025-11-10 21:56
    Function calling support
  • Issue#80sophiahhuang2025-11-04 21:48
    Clarification on Frame Indexing and Clip Segmentation in COSMOS-REASON1
  • Issue#79sophiahhuang2025-11-04 21:48
    Build Failure Report: Compiling vLLM with PyTorch on NVIDIA AArch64 Platform
  • Issue#79sophiahhuang2025-11-04 21:48
    Build Failure Report: Compiling vLLM with PyTorch on NVIDIA AArch64 Platform
  • Issue#76sophiahhuang2025-11-04 21:47
    Cosmos-Reason1-7B, Blackwell PRO 6000, Invalid Argument Error on vLLM 0.10.2
  • Issue#76sophiahhuang2025-11-04 21:47
    Cosmos-Reason1-7B, Blackwell PRO 6000, Invalid Argument Error on vLLM 0.10.2
  • Issue#80wwccww2025-11-04 03:03
    Clarification on Frame Indexing and Clip Segmentation in COSMOS-REASON1
  • Issue#73yangyioryy2025-10-22 02:06
    How are answers in cosmos-reason1-benchmark labeled?
  • Pull request#74kowyo2025-10-05 16:41
  • Issue#73yangyioryy2025-10-05 09:35
    How are answers in cosmos-reason1-benchmark labeled?
  • Issue#72sophiahhuang2025-10-03 23:46
    Cosmos-Reason1-7B on vLLM only processes the first image in a multi-image Chat Completions request from Hugging Face Discussions
  • Issue comment#65pjannaty2025-10-01 23:13
    Inference fails (CUDA invalid argument in flash-attention)
  • Issue#64pjannaty2025-10-01 23:12
    KeyError in DataLoader worker when Training on bridgev2 dataset
  • Issue comment#67shunzh2025-10-01 22:55
    bridgev2 video data
  • Issue comment#64shunzh2025-10-01 22:46
    KeyError in DataLoader worker when Training on bridgev2 dataset

Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 441 stars here means stars gained during the window, not the repo's star count.