Benchmark Almanac

EgoBlind

GPT-4o Mar 2025 59.3% leads since March 2025
Not in model cards

Models answer questions from blind users using only video recorded before each question. GPT-4o mini checks each answer against human references.

The authors scored 16 models on 2,565 questions from 656 videos. GPT-4o scored 59.3% and people scored 87.4%. VISTA tested ten models with a different judge.

Released
March 2025
Kind
Video understanding
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
Items from
Real work and use
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Gemini 2.5 Flash Sep 2025 #112 on the capability index
Last model added
September 2025
Scores refreshed
November 20th, 2026
16
Entries on its own leaderboard, 7 models we track

GPT-4o leads with 59.3% among models released by September 2025 GPT-4o leads with 59.3% among models released by September 2025

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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. This view shows only models we track, each at its best.

The newest model with a score here came out in September 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 GPT-4o Mar 2025, Resolution ori, #F 32 59.3%
  2. 2 Gemini 2.5 Flash Sep 2025, Resolution ori, #F 32 56%
  3. 3 InternVL2.5-26B InternLM2.5-Chat-20B, Resolution 448², #F 8 55%
  4. 4 LLaVA-OV Qwen2.7B, Resolution 384², #F 16 54.5%
  5. 5 InternVL2.5-8B InternLM2.5-Chat-7B, Resolution 448², #F 8 53.5%
  6. 6 Gemini 1.5 Flash Sep 2024, Resolution ori, #F 32 51.8%
  7. 7 LLaVA-Video Qwen2-7B, Resolution 384², #F 1fps 51.5%
  8. 8 Gemini 2.0 Flash Feb 2025, Resolution ori, #F 32 49.9%
  9. 9 Video-LLaMA3 Qwen2.5-7B, Resolution ori, #F 1fps 49.2%
  10. 10 VILA1.5 LLaMA3-8B, Resolution 336², #F 8 48.2%
  11. 11 Qwen2.5-VL Qwen2.5-7B, Resolution ori, #F 1fps 45.5%
  12. 12 MiniCPM-V-2_6 Qwen2-7B, Resolution 384², #F 1fps 40.7%
  13. 13 CogVLM2-Video LLaMA3-8B, Resolution 224², #F 24 40.3%
  14. 14 LLaMA-VID Vicuna-7B, Resolution 224², #F 1fps 39.1%
  15. 15 Video-LLaVA Vicuna-7B, Resolution 224², #F 8 38.1%
  16. 16 ShareGPT4Video LLaMA3-8B, Resolution ori, #F 2fps 32.9%

Source: Official EgoBlind leaderboard

#EntryScoreSettingDateLinkSource
1 GPT-4o Mar 2025 59.3% Resolution ori, #F 32 Mar 2025 Official leaderboard
2 Gemini 2.5 Flash Sep 2025 56% Resolution ori, #F 32 Sep 2025 Official leaderboard
3 InternVL2.5-26B 55% InternLM2.5-Chat-20B, Resolution 448², #F 8 Dec 2024 github.com Official leaderboard
4 LLaVA-OV 54.5% Qwen2.7B, Resolution 384², #F 16 github.com Official leaderboard
5 InternVL2.5-8B 53.5% InternLM2.5-Chat-7B, Resolution 448², #F 8 Dec 2024 github.com Official leaderboard
6 Gemini 1.5 Flash Sep 2024 51.8% Resolution ori, #F 32 Aug 2024 Official leaderboard
7 LLaVA-Video 51.5% Qwen2-7B, Resolution 384², #F 1fps github.com Official leaderboard
8 Gemini 2.0 Flash Feb 2025 49.9% Resolution ori, #F 32 Feb 2025 Official leaderboard
9 Video-LLaMA3 49.2% Qwen2.5-7B, Resolution ori, #F 1fps github.com Official leaderboard
10 VILA1.5 48.2% LLaMA3-8B, Resolution 336², #F 8 github.com Official leaderboard
11 Qwen2.5-VL 45.5% Qwen2.5-7B, Resolution ori, #F 1fps github.com Official leaderboard
12 MiniCPM-V-2_6 40.7% Qwen2-7B, Resolution 384², #F 1fps Aug 2024 github.com Official leaderboard
13 CogVLM2-Video 40.3% LLaMA3-8B, Resolution 224², #F 24 github.com Official leaderboard
14 LLaMA-VID 39.1% Vicuna-7B, Resolution 224², #F 1fps github.com Official leaderboard
15 Video-LLaVA 38.1% Vicuna-7B, Resolution 224², #F 8 github.com Official leaderboard
16 ShareGPT4Video 32.9% LLaMA3-8B, Resolution ori, #F 2fps github.com Official leaderboard
#ModelScoreSettingDateLinkSource
1 GPT-4o Mar 2025 59.3% Resolution ori, #F 32 Mar 2025 Official leaderboard
2 Gemini 2.5 Flash Sep 2025 56% Resolution ori, #F 32 Sep 2025 Official leaderboard
3 InternVL2.5-26B 55% InternLM2.5-Chat-20B, Resolution 448², #F 8 Dec 2024 github.com Official leaderboard
4 InternVL2.5-8B 53.5% InternLM2.5-Chat-7B, Resolution 448², #F 8 Dec 2024 github.com Official leaderboard
5 Gemini 1.5 Flash Sep 2024 51.8% Resolution ori, #F 32 Aug 2024 Official leaderboard
6 Gemini 2.0 Flash Feb 2025 49.9% Resolution ori, #F 32 Feb 2025 Official leaderboard
7 MiniCPM-V-2_6 40.7% Qwen2-7B, Resolution 384², #F 1fps Aug 2024 github.com Official leaderboard

Rankings by part

We read 6 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Tool Use, Information and Navigation. GPT-4o leads 4 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Tool Use

7 models we track have a score here.

  1. 1 Gemini 1.5 Flash Sep 2024 ↑4 72.5%
  2. 2 InternVL2.5-26B ↑1 72.5%
  3. 3 Gemini 2.5 Flash Sep 2025 ↓1 67.1%
  4. 4 GPT-4o Mar 2025 ↓3 66.4%
  5. 5 InternVL2.5-8B ↓1 61.1%
  6. 6 Gemini 2.0 Flash Feb 2025 61.1%
  7. 7 MiniCPM-V-2_6 53.7%

2 models have held number one on the official leaderboard since its release in March 2025

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

Shanghai AI Lab OpenAI
Own leaderboard Not tested since Sep 2025
GPT-4o 59.3%

Source: Official EgoBlind leaderboard

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
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-4o InternVL2.5-26B Gemini 1.5 Flash MiniCPM-V-2_6 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-4o Gemini 1.5 Flash

Source: Official EgoBlind 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 14 months ago, in September 2025, for Gemini 2.5 Flash. 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 official leaderboard (16 entries).
  • Pass Models scored 7 models we track, plus 9 other entries on its leaderboard.
  • Warning Newest score The newest came 14 months ago, in September 2025, for Gemini 2.5 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 2.5 Flash (#112).
  • Pass Headroom The best is 59.3%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 7 rows for models we track and 9 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