Video understanding is evaluated across short, medium and long clips through human-annotated questions.
56 model results
One row per model and compatible score measure. Mean = equal-weight arithmetic average of imported reports. Distinct configurations reported for the same model in a source table remain separate. Reports can overlap; the range is not a confidence interval. Unbundle to inspect each observation.
Published observations span 0.7 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
Published observations span 0.2 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#2 of 56 in this displayed comparison; 2.15 pp below the leader.
Published observations span 3.8 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#3 of 56 in this displayed comparison; 2.45 pp below the leader.
Published observations span 4.2 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#4 of 56 in this displayed comparison; 2.75 pp below the leader.
Published observations span 4.8 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#5 of 56 in this displayed comparison; 6.65 pp below the leader.
Published observations span 3.8 percentage points. Configuration and source differences contribute to this spread; it is not measured run-to-run uncertainty.
#7 of 56 in this displayed comparison; 8.55 pp below the leader.
#10 of 56 in this displayed comparison; 14.55 pp below the leader.
Source notes
This result is for the Qwen2.5-VL-72B-Instruct model augmented by the AdaReTaKe system.
InternVL2 5-78B →Shanghai AI Lab,SenseTime,Tsinghua University,Nanjing University,Fudan University,Chinese University of Hong Kong (CUHK),Shanghai Jiao Tong University
Earlier-2026 baseline comparison from the official Qwen model card. Original model labels and score formatting are retained. Language and vision tables use different Qwen reference models. Tools, image inputs, agent scaffolds and task subsets can differ; follow the model card’s evaluation details. Video-MME subtitle settings and HLE tool settings stay in separate rows. Original MCP-Mark results predate MCPMark Verified.
VideoMME(w/o sub.) · Qwen report · 2026-10-26
Earlier-2026 baseline comparison from the official Qwen model card. Original model labels and score formatting are retained. Language and vision tables use different Qwen reference models. Tools, image inputs, agent scaffolds and task subsets can differ; follow the model card’s evaluation details. Video-MME subtitle settings and HLE tool settings stay in separate rows. Original MCP-Mark results predate MCPMark Verified.
VideoMME · Overall (no subtitles) · % · 2026-10-26
Epoch AI public data snapshot, CC BY 4.0. Externally collected result; original evaluation provider is linked. CSV: video_mme_external.csv; score column: Overall (no subtitles); fraction scale: 1. Model version is retained in configuration. Original source authors retain credit and any additional licensing terms.
VideoMME · Overall (no subtitles) · % · 2026-10-26
Epoch AI public data snapshot, CC BY 4.0. Externally collected result; original evaluation provider is linked. CSV: video_mme_external.csv; score column: Overall (no subtitles); fraction scale: 1. Model version is retained in configuration. Original source authors retain credit and any additional licensing terms.
TRACE THE SCORE · Per-model notes use published breakdowns where available.
Score vs. release date
Epoch AI · 50 observations
VideoMME · Overall (no subtitles) · % · higher is better
Dates are model release dates recorded by Epoch AI, not evaluation dates. Each point retains its published score and configuration. Separate releases are not a matched longitudinal experiment. Missing dates are omitted. Explore all reported results ↓
The dashed line traces the best observed score through the model release dates. It describes this selection of reports, with their original configurations. It does not estimate unmeasured models or costs.
View plotted values and sources (50)
Video-MME · VideoMME · Overall (no subtitles) · %. Published measurements in their original score units.
These model reports include this benchmark. Their harnesses, task subsets, scoring policies, and resource budgets can differ; follow each report’s evaluation notes before joining the numbers.
Video understanding is evaluated across short, medium and long clips through human-annotated questions.
How to read the result
Report frame sampling, video resolution and subtitle access. Models can receive materially different evidence under different frame budgets.
Limits and comparison traps
Scores with subtitles, audio or extra frames need separate labels. This evaluates video input to a language model, not video generation.
Benchmark reference notes
Full-spectrum multi-modal evaluation of video LLMs across short, medium, and long videos (900 videos, 2,700 QA pairs).
Epoch data and evaluation setting
Epoch AI publishes this benchmark in its capability collection. Imported rows retain the published score column and model configuration. Original authors and evaluation providers remain credited in the result references.
Epoch attribution and score coverage
Data is attributed to Epoch AI under CC BY 4.0, with original evaluation providers credited in the linked source snapshots. Scores are imported only where a published mapping identifies the field and its scale; missing or incompatible records are omitted rather than assigned zero.
Where this fits in an evaluation suite
Ask what the task actually requires: recall, deduction, search, or a mixture. Separate closed-book reasoning from retrieval and tool-assisted problem solving.
Start with a capability hypothesis, then choose a second test with a different failure mode. Keep the checkpoint, reasoning effort, agent tools, sampling policy, and scoring procedure fixed when measuring a training intervention. Record failures and missing outputs in the denominator.
A reproducible comparison
Pin the benchmark release, task split, model configuration, tools, sampling budget, and evaluation harness described by the primary source.
Publish the exact run configuration, item coverage, score units, and available uncertainty. If a source supplies only an aggregate or a single run, preserve that limit. A new model name or a rounded score alone does not establish a capability improvement.