Benchmark Almanac

MultiRef-Compass

Gemini-Omni 0.24 tops its leaderboard
Not in model cards

Models generate audio-video content from several visual references and text instructions across 350 samples. Fourteen metrics assess quality, reference fidelity, audio-visual consistency, and instruction following. The evaluator combines automatic measurements with multimodal model judgments.

The public dataset includes 300 main samples and 50 challenge samples. Input references are separate from generated model outputs. Metric backends and judge settings affect comparisons.

Released
July 2026
Kind
Video generation
Level
Everyday
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Images → Video, Audio
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone

In the authors’ words

Multi-reference-to-audio-video (MR2AV) generation aims to generate coherent audio-video content conditioned on multiple references and textual instructions. Existing benchmarks mainly focus on text-driven generation, single-reference subject preservation, or isolated audio-video alignment, leaving the emerging MR2AV setting largely unexplored. Compared with these settings, MR2AV requires models to jointly reason over multiple references while generating synchronized visual and audio content. Models must not only preserve each reference faithfully but also correctly bind and compose multiple referenced entities into coherent audio-visual events. To address this gap, we introduce MultiRef-Compass, a unified benchmark for MR2AV generation. It comprises 350 carefully curated samples constructed through a scalable and controllable asset-composition pipeline, covering multi-view subject preservation, multi-entity binding, and human-object-scene composition. To provide interpretable assessment, MultiRef-Compass defines an evaluation protocol with four dimensions: Basic Quality, Reference Consistency, Audio-Visual Consistency, and Instruction Following, using 14 sub-metrics. MultiRef-Compass integrates automatic metrics with a rejudging-enhanced MLLM-as-a-Judge framework, enabling scalable and auditable evaluation of both perceptual fidelity and reference-conditioned composition. Extensive experiments on eight representative MR2AV systems reveal substantial room for improvement across multiple evaluation dimensions, underscoring the need for a comprehensive benchmark and positioning MultiRef-Compass as a foundation for future MR2AV research.

Abstract of MultiRef-Compass: Towards Comprehensive Evaluation of Multi-Reference-to-Audio-Video Generation (arXiv:2607.14189, CC0)
8
Entries in its paper, 0 models we track

Gemini-Omni tops the leaderboard with 0.24

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper reports separate metrics on different scales. The headline is visual technical quality (VTQ), the first metric in Table 2. 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.

  1. 1 Gemini-Omni 0.24
  2. 2 Seedance 2.0 0.22
  3. 3 SkyReels-V3 0.21
  4. 4 Vidu Q3-Mix 0.20
  5. 5 HappyHouse 1.1 0.20
  6. 6 Kling 3.0 0.19
  7. 7 Wan 2.7 0.18
  8. 8 Wan2.1-VACE 0.16

Source: The MultiRef-Compass paper

#EntryScoreSettingDateSource
1 Gemini-Omni closed weights 0.24 Official leaderboard
2 Seedance 2.0 closed weights 0.22 Official leaderboard
3 SkyReels-V3 open weights 0.21 Official leaderboard
4 Vidu Q3-Mix closed weights 0.20 Official leaderboard
5 HappyHouse 1.1 closed weights 0.20 Official leaderboard
6 Kling 3.0 closed weights 0.19 Official leaderboard
7 Wan 2.7 closed weights 0.18 Official leaderboard
8 Wan2.1-VACE open weights 0.16 Official 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

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the benchmark’s paper (8 entries).
  • Pass Models scored 8 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 8 entries at the last read, none a model we track.
  • Pass Last read November 20th, 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