aiqbench analysis
This paper introduces Watch-Think-Interact (WTI), a closed-loop framework that solves the challenge of streaming video reasoning by dynamically managing context budgets without replaying full history. By training with Stream-GDPO on 82K timed questions, WTI achieves state-of-the-art performance on benchmark datasets while maintaining compact natural-language memory entries for selective recall. The work advances multi-turn interaction capabilities in vision-language models, offering a scalable approach to handling asynchronous queries under strict context constraints.
Watch-Think-Interact: Bootstrapping Long-Horizon Multi-Turn Streaming Video Reasoning with Reinforcement Learning
arXiv:2609.37035v1 Announce Type: new Abstract: Streaming video assistance requires models to answer asynchronous questions from an observed prefix under a fixed context budget. Existing approaches model response timing or compress history, but an online state formed before future questions are known can omit visual details before later questions reveal their relevance; the retained state alone cannot recover them. We introduce Watch-Think-Interact (WTI), a closed-loop framework for multi-question streaming video reasoning. WTI maintains compact natural-language memory entries tagged with source-video time ranges; these entries support direct reasoning when sufficient and otherwise anchor selective recall of finer visual evidence. For each question, WTI answers when current context and memory suffice, continues watching when required evidence has not appeared, or recalls a relevant past interval and decides again after incorporating the returned chunks, without replaying the full observed history. To train this behavior, we construct WTI-82K, comprising 82,335 timed questions across 4,812 causally aligned trajectories, and develop Stream-GDPO to optimize complete multi-question streaming rollouts using trajectory-level feedback for response timing, source-video recall, and memory updates. WTI achieves state-of-the-art aggregate performance among the compared open-source streaming baselines, reaching 83.3% on StreamingBench and 73.6% weighted overall accuracy on OVO-Bench.
Abstract:Streaming video assistance requires models to answer asynchronous questions from an observed prefix under a fixed context budget. Existing approaches model response timing or compress history, but an online state formed before future questions are known can omit visual details before later questions reveal their relevance; the retained state alone cannot recover them. We introduce Watch-Think-Interact (WTI), a closed-loop framework for multi-question streaming video reasoning. WTI maintains compact natural-language memory entries tagged with source-video time ranges; these entries support direct reasoning when sufficient and otherwise anchor selective recall of finer visual evidence. For each question, WTI answers when current context and memory suffice, continues watching when required evidence has not appeared, or recalls a relevant past interval and decides again after incorporating the returned chunks, without replaying the full observed history. To train this behavior, we construct WTI-82K, comprising 82,335 timed questions across 4,812 causally aligned trajectories, and develop Stream-GDPO to optimize complete multi-question streaming rollouts using trajectory-level feedback for response timing, source-video recall, and memory updates. WTI achieves state-of-the-art aggregate performance among the compared open-source streaming baselines, reaching 83.3% on StreamingBench and 73.6% weighted overall accuracy on OVO-Bench.
Artificial Intelligence (cs.AI)
arXiv:2609.37035 [cs.AI]
(or arXiv:2609.37035v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.37035 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Full text under the publisher's open licence · 283 words
Related coverage
What AI gets wrong and what failure teaches us
Microsoft Research · Oct 6, 2026
Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUC
Latent Space · Sep 16, 2026
State of the Field: AI for Epistemics and Coordination
LessWrong · Oct 2, 2026
Tariffs didn’t bring manufacturing jobs back to the US
The Verge · Jul 27, 2026