Benchmark Almanac

FortisAVQA

MAVEN 74.7% tops its leaderboard VITA 48.7% leads the models we track
Not in model cards

Models answer questions about sounds and visible events in music videos. Accuracy is reported separately for frequent and rare question distributions after the original questions are rephrased.

The 211,572-question test set derives from MUSIC-AVQA. A sampled subset reduces evaluation costs for large models.

Released
April 2025
Kind
Video understanding
Level
Everyday
Scored
Checked automatically
Answers
Short answer
Tasks
All public
Measures
Capability
Subject
Arts
Language
English
In and out
Text, Video, Audio → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
Last model added
August 2024
Scores refreshed
November 19th, 2026

In the authors’ words

Audio-Visual Question Answering (AVQA) is a challenging multimodal reasoning task requiring intelligent systems to answer natural language queries based on paired audio-video inputs accurately. However, existing AVQA approaches often suffer from overfitting to dataset biases, leading to poor robustness. Moreover, current datasets may not effectively diagnose these methods. To address these challenges, we first introduce a novel dataset, FortisAVQA, constructed in two stages: (1) rephrasing questions in the test split of the public MUSIC-AVQA dataset and (2) introducing distribution shifts across questions. The first stage expands the test space with greater diversity, while the second enables a refined robustness evaluation across rare, frequent, and overall question distributions. Second, we introduce a robust Multimodal Audio-Visual Epistemic Network (MAVEN) that leverages a multifaceted cycle collaborative debiasing strategy to mitigate bias learning. Experimental results demonstrate that our architecture achieves state-of-the-art performance on FortisAVQA, with a notable improvement of 7.81%. Extensive ablation studies on both datasets validate the effectiveness of our debiasing components. Additionally, our evaluation reveals the limited robustness of existing multimodal QA methods. We also verify the plug-and-play capability of our strategy by integrating it with various baseline models across both datasets. Our dataset and code are available at https://github.com/reml-group/fortisavqa.

Abstract of FortisAVQA and MAVEN: a Benchmark Dataset and Debiasing Framework for Robust Multimodal Reasoning (arXiv:2504.00487, CC0)
13
Entries in its paper, 1 model we track

MAVEN tops the leaderboard with 74.7% VITA leads with 48.7%

Best score per entry from the benchmark's paper, read on November 19th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is VITA, at 48.7%. This view shows only models we track, each at its best.

  1. 1 MAVEN MCCD strategy: ✓ 74.7%
  2. 2 VideoLLaMA 2 MCCD strategy: ✗ 66.8%
  3. 3 MCAN MCCD strategy: ✓ 57.5%
  4. 4 Qwen2.5-VL MCCD strategy: ✗ 57.1%
  5. 5 LAVisH MCCD strategy: ✓ 56.2%
  6. 6 GPT 4o (snapshot unspecified) MCCD strategy: ✗ 56.1%
  7. 7 GRU MCCD strategy: ✓ 55.8%
  8. 8 CONVLSTM MCCD strategy: ✓ 55.8%
  9. 9 FCNLSTM MCCD strategy: ✗ 54.1%
  10. 10 PSAC MCCD strategy: ✗ 51.8%
  11. 11 VITA MCCD strategy: ✗ 48.7%
  12. 12 LAViT MCCD strategy: ✓ 48.1%
  13. 13 HCRN MCCD strategy: ✓ 42.8%

Source: The FortisAVQA paper

Source: The FortisAVQA paper

#EntryScoreSettingDateSource
1 MAVEN 74.7% MCCD strategy: ✓ Official leaderboard
2 VideoLLaMA 2 66.8% MCCD strategy: ✗ Official leaderboard
3 MCAN 57.5% MCCD strategy: ✓ Official leaderboard
4 Qwen2.5-VL 57.1% MCCD strategy: ✗ Official leaderboard
5 LAVisH 56.2% MCCD strategy: ✓ Official leaderboard
6 GPT 4o (snapshot unspecified) 56.1% MCCD strategy: ✗ Official leaderboard
7 GRU 55.8% MCCD strategy: ✓ Official leaderboard
8 CONVLSTM 55.8% MCCD strategy: ✓ Official leaderboard
9 FCNLSTM 54.1% MCCD strategy: ✗ Official leaderboard
10 PSAC 51.8% MCCD strategy: ✗ Official leaderboard
11 VITA 48.7% MCCD strategy: ✗ Aug 2024 Official leaderboard
12 LAViT 48.1% MCCD strategy: ✓ Official leaderboard
13 HCRN 42.8% MCCD strategy: ✓ Official leaderboard
#ModelScoreSettingDateSource
1 VITA 48.7% MCCD strategy: ✗ Aug 2024 Official leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

In the benchmark’s paper
0% 20% 40% 60% 80% 100% Near the ceiling (85%) VITA 0% 20% 40% 60% 80% 100% Near the ceiling (85%) VITA

Source: Official FortisAVQA leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Audio and video

All 695 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in August 2024, for VITA. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (13 entries).
  • Pass Models scored 1 model we track, plus 12 other entries on its leaderboard.
  • Warning Newest score The newest came 2 years ago, in August 2024, for VITA.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 48.7%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 1 row for models we track and 12 other entries at the last read.
  • Pass Last read November 19th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check