Pith Number

pith:JYME5X3P

pith:2026:JYME5X3PIZH3HHW2Q6NRPWEO5Z

not attested not anchored not stored refs resolved

How Far Are Video Models from True Multimodal Reasoning?

Daoan Zhang, Dezhi YU, Jianhui Wei, Jie Tan, Songtao Jiang, Wei Xu, Xiaotian Zhang, Yan Zhang, Yichen Li, Yuan Wang, Ziyi Chen, Zuozhu Liu

State-of-the-art video models handle basic understanding but fail on logically grounded and interactive video generation tasks.

arxiv:2604.19193 v1 · 2026-04-21 · cs.CV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{JYME5X3PIZH3HHW2Q6NRPWEO5Z}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

while state-of-the-art (SOTA) video models, such as Seedance 2.0, demonstrate competence on certain understanding and reasoning subtasks, they fall substantially short with logically grounded and interactive generation tasks (achieving success rates <25% and ~0%, respectively)

C2weakest assumption

That the manually annotated CLVG-Bench tasks and the Adaptive Video Evaluator (AVE) accurately capture and measure 'true multimodal reasoning' in a manner that aligns with human expert perception without bias or incompleteness.

C3one line summary

Current video models succeed on basic understanding but achieve under 25% success on logically grounded generation and near 0% on interactive generation, exposing gaps in multimodal reasoning.

References

103 extracted · 103 resolved · 28 Pith anchors

[1] GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning 2025 · arXiv:2507.19457

[2] Oxford University Press (1984) 6 1984

[3] VideoPhy: Evaluating Physical Commonsense for Video Generation 2024 · arXiv:2406.03520

[4] Videophy-2: A challenging action-centric physical commonsense evaluation in video generation 2025

[5] Bordwell, D., Thompson, K., Smith, J.: Film art: An introduction, vol. 7. McGraw- Hill New York (2008) 6 2008

Cited by

1 paper in Pith

VLMs are Good Teachers for Video Reasoning via Adaptive Test-Time Optimization

Receipt and verification

First computed	2026-06-26T01:15:18.907568Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

4e184edf6f464fb39eda879b17d88eee7b5845e06b3a0ead8c45f9eb3771784f

Aliases

arxiv: 2604.19193 · arxiv_version: 2604.19193v1 · doi: 10.48550/arxiv.2604.19193 · pith_short_12: JYME5X3PIZH3 · pith_short_16: JYME5X3PIZH3HHW2 · pith_short_8: JYME5X3P

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/JYME5X3PIZH3HHW2Q6NRPWEO5Z \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 4e184edf6f464fb39eda879b17d88eee7b5845e06b3a0ead8c45f9eb3771784f

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "fdf92f640d25835c64e3b0925c955c67568a9c4430d921efe72886ad8b4d3d45",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-04-21T08:04:02Z",
    "title_canon_sha256": "7cc9f0fbdad3f022736064eafce86de064a9f35eee9f802069c85b2590541a18"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.19193",
    "kind": "arxiv",
    "version": 1
  }
}