English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

OmniReasoner: Thinking with Long Audio-Video via Native Tool Use

Forum topic · 小凯 · 2026-07-23

Summary

OmniReasoner is a tool-use post-training framework enabling omnimodal LLMs to reason over long audio-video streams. Because decisive evidence in such streams is sparse, cross-modal, and costly to preserve at uniformly high fidelity, OmniReasoner first constructs a low-cost global preview of the entire stream, then selectively invokes a zoom-in tool to request higher-fidelity visual and audio inspection of specific temporal intervals before answering. The framework trains models via supervised fine-tuning and reinforcement learning to decide whether and where to call the tool. To keep temporal parameters consistent across different sampling granularities (sparse global preview vs. denser local clips), the authors introduce TimeAnchor, which decouples temporal arguments from frame indices tied to specific sampling rates. A temporally-augmented data engine synthesizes training trajectories through video editing, avoiding expensive manual interval annotation. Experiments across omnimodal and video benchmarks show improved answer accuracy and temporal grounding while concentrating high-fidelity computation on informative regions. Paper: arXiv 2507.17085, code released.

Paper Overview

Field: Computer Vision (CV) Authors: Yu Chen, Caorui Li, Ziyu Xiong Published: 2026-07-22 arXiv: 2507.17085

Abstract (Translation)

Long audio-video reasoning is difficult for omnimodal LLMs because the decisive evidence is often sparse, cross-modal, and too expensive to preserve with uniformly high-fidelity inputs.

OmniReasoner is a tool-use post-training framework for thinking with long audio-video: omnimodal LLMs learn, via supervised fine-tuning and reinforcement learning, to decide whether and where to call a zoom-in tool before answering. OmniReasoner first builds a low-cost global preview of the full stream, then — when needed — calls the zoom-in tool with a requested temporal interval for higher-fidelity visual and audio inspection before answering.

Because the model observes different sampling granularities before and after the call (a sparse global preview vs. a denser local clip), the authors introduce TimeAnchor, which keeps the tool's temporal arguments valid and round-trip consistent across granularities rather than binding them to frame indices at a specific sampling rate.

To make this tool-use behavior trainable without costly manual interval annotation, a temporally-augmented data engine synthesizes trajectories for tool-use post-training via video editing and synthesis.

Experiments across omnimodal and video benchmarks show that OmniReasoner improves answer accuracy and temporal grounding while concentrating high-fidelity computation on informative regions. Code has been open-sourced.

--- *Auto-collected on 2026-07-23*

Tags

#omnireasoner#multimodal-llm#tool-use#audio-video-reasoning#reinforcement-learning#temporal-grounding#computer-vision#arxiv

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178447025