Benchmark Almanac

PLM-VideoBench

PLM-8B 55.6 tops its leaderboard Gemini 2.0 Flash Feb 2025 42.5 leads the models we track
Not in model cards

Models answer questions about activities, describe events involving a particular person or object, and locate those events in time. Meta also tests first-person video understanding.

Meta averages five task scores, including automatic checks and model grading.

Made by
Meta FAIR
Released
April 2025
Kind
Video captioning
Level
Everyday
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Video → Text
Interaction
Single turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Gemini 2.0 Flash Feb 2025 #147 on the capability index
Last model added
February 2025
Scores refreshed
November 20th, 2026
9
Entries in its paper, 6 models we track

PLM-8B tops the leaderboard with 55.6 Gemini 2.0 Flash leads with 42.5 among models released by February 2025

Best score per entry from the benchmark's paper, read on November 20th, 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 Gemini 2.0 Flash, at 42.5. This view shows only models we track, each at its best.

The newest model with a score here came out in February 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 PLM-8B 55.6
  2. 2 GPT-4o (snapshot unspecified) 51.6
  3. 3 Gemini 1.5 Pro (snapshot unspecified) 44.0
  4. 4 Gemini 2.0 Flash Feb 2025 42.5
  5. 5 InternVL2.5-8B 38.5
  6. 6 Qwen2-VL-7B-Instruct 35.3
  7. 7 InternVL2-8B 35.0
  8. 8 Qwen2.5-VL-7B 34.8
  9. 9 LLaVA-OneVision 7B 32.0

Source: The PLM-VideoBench paper

#EntryScoreSettingDateSource
1 PLM-8B 55.6 Official leaderboard
2 GPT-4o (snapshot unspecified) 51.6 Official leaderboard
3 Gemini 1.5 Pro (snapshot unspecified) 44.0 Official leaderboard
4 Gemini 2.0 Flash Feb 2025 42.5 Feb 2025 Official leaderboard
5 InternVL2.5-8B 38.5 Dec 2024 Official leaderboard
6 Qwen2-VL-7B-Instruct 35.3 Aug 2024 Official leaderboard
7 InternVL2-8B 35.0 Jul 2024 Official leaderboard
8 Qwen2.5-VL-7B 34.8 Jan 2025 Official leaderboard
9 LLaVA-OneVision 7B 32.0 Aug 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Gemini 2.0 Flash Feb 2025 42.5 Feb 2025 Official leaderboard
2 InternVL2.5-8B 38.5 Dec 2024 Official leaderboard
3 Qwen2-VL-7B-Instruct 35.3 Aug 2024 Official leaderboard
4 InternVL2-8B 35.0 Jul 2024 Official leaderboard
5 Qwen2.5-VL-7B 34.8 Jan 2025 Official leaderboard
6 LLaVA-OneVision 7B 32.0 Aug 2024 Official leaderboard

Rankings by part

We read 5 parts from the benchmark’s paper, each ranking models on one part of the test, such as FGQA MBAcc, SGQA accuracy and RDCap SODA. Gemini 2.0 Flash leads 4 of them, and InternVL2.5-8B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

FGQA MBAcc

6 models we track have a score here.

  1. 1 Gemini 2.0 Flash Feb 2025 58.7
  2. 2 InternVL2.5-8B 53.7
  3. 3 Qwen2.5-VL-7B ↑2 49.8
  4. 4 Qwen2-VL-7B-Instruct ↓1 49.2
  5. 5 InternVL2-8B ↓1 47.7
  6. 6 LLaVA-OneVision 7B 40.2

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Benchmark paper Not tested since Apr 2025

Source: The PLM-VideoBench paper

Scores over time

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

Plot with other benchmarks
In the benchmark’s paper
30.0 32.5 35.0 37.5 40.0 42.5 2025 30.0 32.5 35.0 37.5 40.0 42.5 2025

Source: Official PLM-VideoBench 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

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 21 months ago, in February 2025, for Gemini 2.0 Flash. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (9 entries).
  • Pass Models scored 6 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 21 months ago, in February 2025, for Gemini 2.0 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 2.0 Flash (#147).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 6 rows for models we track and 3 other entries at the last read.
  • 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