Benchmark Almanac

OpenLMM Subjective Leaderboard

GPT-4.1 76.6 leads since April 2025
Not in model cards

Models generate open-ended answers to multimodal tasks. The board reports subjective evaluation results.

The leaderboard aggregates results from component evaluations.

Made by
OpenCompass
Released
April 2025
Kind
Composite indices
Level
University
Scored
Settled by events
Answers
Short answer
Tasks
All public
Measures
Capability
In and out
Text, Images → Text
Interaction
Single turn
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Claude 3.7 Sonnet #119 on the capability index
Last model added
April 2025
Scores refreshed
November 20th, 2026
29
Entries on its own leaderboard, 26 models we track

GPT-4.1 leads with 76.6 among models released by April 2025 GPT-4.1 leads with 76.6 among models released by April 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 April 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-4.1 76.6
  2. 2 Qwen2.5-VL-32B 73.8
  3. 3 Qwen2.5-VL-72B 73.7
  4. 4 GPT-4o (1120, detail-high) 72.5
  5. 5 Gemini 2.5 Flash Apr 2025 71.7
  6. 6 Claude 3.7 Sonnet 71.5
  7. 7 Gemini 2.0 Flash Feb 2025 70.1
  8. 8 Claude 3.5 Sonnet 68.9
  9. 9 Grok 2 Vision (1212) 68.9
  10. 10 InternVL3-78B 67.6
  11. 11 GPT-4o (0806, detail-high) 66.6
  12. 12 GPT-4o-mini (0718, detail-high) 65.9
  13. 13 Llama 4 Scout 65.5
  14. 14 InternVL3-38B 64.3
  15. 15 InternVL3-8B 63.7
  16. 16 Ovis2-34B 63.2
  17. 17 InternVL2_5-78B 62.7
  18. 18 Pixtral 12B 60.4
  19. 19 InternVL2.5-38B 59.9
  20. 20 Qwen2.5-VL-7B 59.7
  21. 21 InternVL2.5-8B 55.7
  22. 22 LLaVA-OneVision 72B 52.0
  23. 23 NVLM-D-72B 49.8
  24. 24 Llama 3.2 11B 48.1
  25. 25 DeepSeek-VL2 48.1
  26. 26 MiniCPM-o-2_6 46.8
  27. 27 LLaVA-OneVision 7B 44.1
  28. 28 Phi-4-multimodal 42.4
  29. 29 Janus-Pro-7B 36.3

Source: Official OpenLMM Subjective Leaderboard leaderboard

#EntryScoreSettingReleasedRun onLinkSource
1 GPT-4.1 76.6 Apr 2025 Apr 2025 openai.com Official leaderboard
2 Qwen2.5-VL-32B 73.8 Mar 2025 Apr 2025 huggingface.co Official leaderboard
3 Qwen2.5-VL-72B 73.7 Jan 2025 Apr 2025 huggingface.co Official leaderboard
4 GPT-4o (1120, detail-high) 72.5 Apr 2025 openai.com Official leaderboard
5 Gemini 2.5 Flash Apr 2025 71.7 Apr 2025 Apr 2025 deepmind.google Official leaderboard
6 Claude 3.7 Sonnet 71.5 Feb 2025 Apr 2025 docs.anthropic.com Official leaderboard
7 Gemini 2.0 Flash Feb 2025 70.1 Feb 2025 Apr 2025 deepmind.google Official leaderboard
8 Claude 3.5 Sonnet 68.9 Jun 2024 Apr 2025 docs.anthropic.com Official leaderboard
9 Grok 2 Vision (1212) 68.9 Dec 2024 Apr 2025 docs.x.ai Official leaderboard
10 InternVL3-78B 67.6 Apr 2025 Apr 2025 huggingface.co Official leaderboard
11 GPT-4o (0806, detail-high) 66.6 Apr 2025 openai.com Official leaderboard
12 GPT-4o-mini (0718, detail-high) 65.9 Apr 2025 openai.com Official leaderboard
13 Llama 4 Scout 65.5 Apr 2025 Apr 2025 huggingface.co Official leaderboard
14 InternVL3-38B 64.3 Apr 2025 Apr 2025 huggingface.co Official leaderboard
15 InternVL3-8B 63.7 Apr 2025 Apr 2025 huggingface.co Official leaderboard
16 Ovis2-34B 63.2 Feb 2025 Apr 2025 huggingface.co Official leaderboard
17 InternVL2_5-78B 62.7 Dec 2024 Apr 2025 huggingface.co Official leaderboard
18 Pixtral 12B 60.4 Sep 2024 Apr 2025 huggingface.co Official leaderboard
19 InternVL2.5-38B 59.9 Dec 2024 Apr 2025 huggingface.co Official leaderboard
20 Qwen2.5-VL-7B 59.7 Jan 2025 Apr 2025 huggingface.co Official leaderboard
21 InternVL2.5-8B 55.7 Dec 2024 Apr 2025 huggingface.co Official leaderboard
22 LLaVA-OneVision 72B 52.0 Aug 2024 Apr 2025 huggingface.co Official leaderboard
23 NVLM-D-72B 49.8 Sep 2024 Apr 2025 huggingface.co Official leaderboard
24 Llama 3.2 11B 48.1 Sep 2024 Apr 2025 huggingface.co Official leaderboard
25 DeepSeek-VL2 48.1 Dec 2024 Apr 2025 huggingface.co Official leaderboard
26 MiniCPM-o-2_6 46.8 Jan 2025 Apr 2025 huggingface.co Official leaderboard
27 LLaVA-OneVision 7B 44.1 Aug 2024 Apr 2025 huggingface.co Official leaderboard
28 Phi-4-multimodal 42.4 Feb 2025 Apr 2025 huggingface.co Official leaderboard
29 Janus-Pro-7B 36.3 Jan 2025 Apr 2025 huggingface.co Official leaderboard
#ModelScoreSettingReleasedRun onLinkSource
1 GPT-4.1 76.6 Apr 2025 Apr 2025 openai.com Official leaderboard
2 Qwen2.5-VL-32B 73.8 Mar 2025 Apr 2025 huggingface.co Official leaderboard
3 Qwen2.5-VL-72B 73.7 Jan 2025 Apr 2025 huggingface.co Official leaderboard
4 Gemini 2.5 Flash Apr 2025 71.7 Apr 2025 Apr 2025 deepmind.google Official leaderboard
5 Claude 3.7 Sonnet 71.5 Feb 2025 Apr 2025 docs.anthropic.com Official leaderboard
6 Gemini 2.0 Flash Feb 2025 70.1 Feb 2025 Apr 2025 deepmind.google Official leaderboard
7 Claude 3.5 Sonnet 68.9 Jun 2024 Apr 2025 docs.anthropic.com Official leaderboard
8 Grok 2 Vision (1212) 68.9 Dec 2024 Apr 2025 docs.x.ai Official leaderboard
9 InternVL3-78B 67.6 Apr 2025 Apr 2025 huggingface.co Official leaderboard
10 Llama 4 Scout 65.5 Apr 2025 Apr 2025 huggingface.co Official leaderboard
11 InternVL3-38B 64.3 Apr 2025 Apr 2025 huggingface.co Official leaderboard
12 InternVL3-8B 63.7 Apr 2025 Apr 2025 huggingface.co Official leaderboard
13 Ovis2-34B 63.2 Feb 2025 Apr 2025 huggingface.co Official leaderboard
14 InternVL2_5-78B 62.7 Dec 2024 Apr 2025 huggingface.co Official leaderboard
15 Pixtral 12B 60.4 Sep 2024 Apr 2025 huggingface.co Official leaderboard
16 InternVL2.5-38B 59.9 Dec 2024 Apr 2025 huggingface.co Official leaderboard
17 Qwen2.5-VL-7B 59.7 Jan 2025 Apr 2025 huggingface.co Official leaderboard
18 InternVL2.5-8B 55.7 Dec 2024 Apr 2025 huggingface.co Official leaderboard
19 LLaVA-OneVision 72B 52.0 Aug 2024 Apr 2025 huggingface.co Official leaderboard
20 NVLM-D-72B 49.8 Sep 2024 Apr 2025 huggingface.co Official leaderboard
21 Llama 3.2 11B 48.1 Sep 2024 Apr 2025 huggingface.co Official leaderboard
22 DeepSeek-VL2 48.1 Dec 2024 Apr 2025 huggingface.co Official leaderboard
23 MiniCPM-o-2_6 46.8 Jan 2025 Apr 2025 huggingface.co Official leaderboard
24 LLaVA-OneVision 7B 44.1 Aug 2024 Apr 2025 huggingface.co Official leaderboard
25 Phi-4-multimodal 42.4 Feb 2025 Apr 2025 huggingface.co Official leaderboard
26 Janus-Pro-7B 36.3 Jan 2025 Apr 2025 huggingface.co Official leaderboard

Rankings by part

We read 7 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Creation_MMBench reward, Creation_MMBench VFS and MIA-Bench. Qwen2.5-VL-32B leads 2 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Creation_MMBench reward

26 models we track have a score here.

  1. 1 Qwen2.5-VL-32B ↑1 6.97
  2. 2 Gemini 2.5 Flash Apr 2025 ↑2 6.61
  3. 3 GPT-4.1 ↓2 5.61
  4. 4 Qwen2.5-VL-72B ↓1 -5.82
  5. 5 InternVL3-78B ↑4 -6.87
  6. 6 Grok 2 Vision (1212) ↑2 -8.86
  7. 7 Gemini 2.0 Flash Feb 2025 ↓1 -8.90
  8. 8 InternVL3-38B ↑3 -8.99
  9. 9 Claude 3.7 Sonnet ↓4 -11.28
  10. 10 InternVL3-8B ↑2 -14.00

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

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

Qwen OpenAI
Own leaderboard Not tested since Apr 2025
GPT-4.1 76.6

Source: Official OpenLMM Subjective Leaderboard 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
30.0 40.0 50.0 60.0 70.0 80.0 2024 2025 30.0 40.0 50.0 60.0 70.0 80.0 2024 2025

Source: Official OpenLMM Subjective Leaderboard 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 Other

All 69 in Other
  • Artificial Analysis Intelligence Index v5 In use Version 5 of Artificial Analysis's index: nine evaluations in four equal categories, most with private test sets, run three times each in its own agent harness. Claude Fable 5.2 64.6
  • Personal Superintelligence Index (Meta) In use Meta's composite of seven personal and knowledge-work evals: Agents' Last Exam, AutomationBench-AA, FocusBench, JobBench, GDP.pdf, VBT and DRACO.
  • Vals Science Index In use Agents are scored on the equal average of BioMysteryBench, MysteryMechanism and Terminal-Bench-Science. GPT-6.2 Astra 68.4%
  • Decision Index Public v0.3 In use Decision models return probabilities over labeled options in public tasks covering knowledge, language, retrieval, tools, and arts. A chance-corrected weighted aggregate combines the task scores. Granite Switch 4.1 30B 51.3
  • SAGE v1.1 (Vals) In use Models grade handwritten undergraduate math and computer science work against rubrics. Scored by class-balanced accuracy of each rubric check against expert grading. Claude Opus 5.5 91.3%
  • MTEB Fading Massive Text Embedding Benchmark: retrieval, clustering, classification across languages and modalities. Qwen3-Embedding-8B 72.2%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 19 months ago, in April 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 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (29 entries).
  • Pass Models scored 26 models we track, plus 3 other entries on its leaderboard.
  • Warning Newest score The newest came 19 months ago, in April 2025, for Gemini 2.5 Flash.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3.7 Sonnet (#119).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 26 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