AI Paper Atlas

Research map / Benchmarks / LongVideoBench

LongVideoBench leaderboard

83 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on LongVideoBench. Best accuracy: 80 (LVAgent, ICCV 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

LongVideoBench · accuracy

57 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
180LVAgentLVAgent: Long Video Understanding by Multi-Round Dynamical Collaboration of MLLM Agents
ICCV 2025 · Table 1
Video-language understanding
272.2A4VLA Multi-Agent Perception-Action Alliance for Efficient Long Video Reasoning
CVPR 2026 · Table 1
Video-language understanding
368.1Qwen3-VL + CLEAR
Qwen3-VL-8B
Diagnosing and Correcting Bias in MLLM for Long Video Understanding
NeurIPS 2026 · Table 1
Video-language understanding
467.8AdaCodec (comparable token budget)
Qwen3-VL-8B
AdaCodec: A Predictive Visual Code for Video MLLMs
NeurIPS 2026 · Table 1
Video-language understanding
567.1ReMem
LLaVA-Video-7B
Reasoning with Memory: A Temporal Granularity-Adaptive Framework for Training-Free Long Video Understanding
ECCV 2026 · Table 1
Video-language understanding
665.4VideoChat-A1
InternVideo2.5-8B
VideoChat-A1: Thinking with Long Videos by Chain-of-Shot Reasoning
AAAI 2026 · Table 1
Video-language understanding
765.2VideoChat-Online
InternVL2-4B
Online Video Understanding: OVBench and VideoChat-Online
CVPR 2025 · Table 3
Video-language understanding
865.1Tempo (8K budget)
Qwen3-VL-2B + Qwen3-LM-4B
Small Vision-Language Models are Smart Compressors for Long Video Understanding
ECCV 2026 · Table 1
Video-language understanding
965Max-U
Qwen2.5-VL-72B-AWQ
VIBE: Annotation-Free Video-to-Text Information Bottleneck Evaluation for TL;DR
NeurIPS 2025 · User Study Results table
Video-language understanding
1064.2DeltaVid-Qwen3-VL-8B
Qwen3-VL-8B
DELTAVID: Enhancing Fine-Grained Spatiotemporal Perception with Cross-Video Differences
NeurIPS 2026 · Table 2
Video-language understanding
47 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

LongVideoBench · accuracy (val)

26 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
167.2InternVideo2.5* + LongVPO Stage2-iter2
InternVideo2.5-8B
LongVPO: From Anchored Cues to Self-Reasoning for Long-Form Video Preference Optimization
NeurIPS 2025 · Main results table
Video-language understanding
266.3ReQuest
Qwen3-VL-8B
ReQuest: Rethinking-based Question-Aware Frame Selection for Long-Form Video QA
ECCV 2026 · Table 1
Video-language understanding
366AdaAlloc (16K)
Qwen3-VL-8B-Instruct
AdaAlloc: Adaptive Visual Token Allocation for Long-Video Question Answering
NeurIPS 2026 · Table 3
Video-language understanding
464.9Qwen2.5-VL-7B w/ APVRAPVR: Hour-Level Long Video Understanding with Adaptive Pivot Visual Information Retrieval
AAAI 2026 · Table 1
Video-language understanding
564.7VideoChat-Flash@448
Qwen2-7B
VideoChat-Flash: Hierarchical Compression for Long-Context Video Modeling
ICLR 2026 · Table 1
Video-language understanding
663.7DynamicSelect
Qwen2.5-VL-7B
Seeing Is Believing: Grounding Long-Video Understanding in Spatio-Temporal Visual Evidence
AAAI 2026 · Table 3
Video-language understanding
763.5ViTL (full)
Qwen2.5-VL
Video-in-the-Loop: Span-Grounded Long Video QA with Interleaved Reasoning
ICML 2026 · Table 7
Video-language understanding
863.3Qwen2.5-VL + EAKV
Qwen2.5-VL-7B
EAKV: An Entropy-Driven Adaptive KV Compression Framework for Long Video Understanding
ICML 2026 · Table 1
Video-language understanding
962.7LLaVA-Video w/ AKS
LLaVA-Video-7B
Adaptive Keyframe Sampling for Long Video Understanding
CVPR 2025 · Table 1
Video-language understanding
1062.5Reflect-R1 (Ours)Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding
ECCV 2026 · Table 1
Video-language understanding
16 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.