Skip to content
AI News HubLIVE
Original source2 min read

Disagreement-Based Cross-Model Routing for Implicit Video Question Answering

Summary

This paper studies implicit video QA and proposes a label-free, training-free inference-time method: triple-sample a native-video model, identify ~20% disagreement questions, and route only that subset to a second model from a different family. It improves AvgAcc by +1.43 on the 1001-question validation set and achieves 82.03 AvgAcc on the 172-question challenge test set.

SourcearXiv Computer VisionAuthor: Durga Sandeep Saluru
Disagreement-Based Cross-Model Routing for Implicit Video Question Answering
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[2606.14723] Disagreement-Based Cross-Model Routing for Implicit Video Question Answering

[Submitted on 31 May 2026]

Title:Disagreement-Based Cross-Model Routing for Implicit Video Question Answering

View a PDF of the paper titled Disagreement-Based Cross-Model Routing for Implicit Video Question Answering, by Durga Sandeep Saluru

View PDF HTML (experimental)

Abstract:We study multiple-choice video question answering on the ImplicitQA benchmark, where the correct answer is never explicitly shown but must be inferred from off-screen events, line-of-sight cues, causal structure, and cross-shot spatial layout. On this benchmark a single frontier video LLM already operates near its accuracy ceiling, and we observe that conventional self-consistency strategies -- majority voting across repeated samples of the same model -- can hurt rather than help, because the model's errors on hard questions are correlated. We propose disagreement-based cross-model routing, a pure inference-time procedure that requires no labels and no training. We triple-sample a native-video model (Gemini 3.1 Pro Preview) at temperature zero, exploit the genuine sample-to-sample variance of its video-processing pipeline to identify the roughly 20% subset of questions where the three samples disagree, and route only that subset to a second model from a different family (Claude Opus 4.8) that consumes uniformly sampled frames with adaptive thinking. On the 1001-question validation set with public ground truth -- our main evaluation -- the method improves AvgAcc by +1.43 over the best single sample of the primary model, with per-category gains concentrated on Motion & Trajectory (+5.49), Inferred Counting (+3.45), and Vertical Spatial Reasoning (+1.82) -- the categories most dependent on cross-shot reference resolution. The same pipeline applied to the held-out 172-question CVPR 2026 ImplicitQA challenge test set achieves 82.03 AvgAcc / 79.71 MacroAvgAcc (+1.81 over the best single sample of the primary model), confirming the validation result on an independent split.

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2606.14723 [cs.CV]

(or arXiv:2606.14723v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2606.14723

arXiv-issued DOI via DataCite

Submission history

From: Durga Sandeep Saluru [view email] [v1] Sun, 31 May 2026 05:56:27 UTC (25 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Disagreement-Based Cross-Model Routing for Implicit Video Question Answering, by Durga Sandeep Saluru

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CV

new | recent | 2026-06

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

InvestorsAdvanced

Key points

  • A single frontier video LLM is near its ceiling on ImplicitQA, and conventional self-consistency voting can hurt due to correlated errors on hard questions
  • Triple-sampling Gemini 3.1 Pro Preview at temperature zero exploits genuine pipeline variance to identify ~20% disagreement questions
  • Routing only that subset to Claude Opus 4.8 with uniform frame sampling and adaptive thinking yields +1.43 AvgAcc on validation
  • Per-category gains concentrate on Motion & Trajectory (+5.49), Inferred Counting (+3.45), and Vertical Spatial Reasoning (+1.82); independent test set shows +1.81 gain

Highlights and analysis are generated automatically and may contain errors. Check the original source.