Benchmark Almanac

MIRAGE-MMST

GPT-4.1 68.0 leads since June 2025
Not in model cards

Models answer growers’ questions using photographs and written descriptions. They identify the cause of a problem and recommend how to manage it.

Three model judges grade identification and management. The standard leaderboard averages those two scores on a 100-point scale.

Released
June 2025
Kind
General visual questions
Level
University
Scored
Graded by a model
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Images → Text
Interaction
Single turn
Items from
Real work and use
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
22
Models scored on its own leaderboard

GPT-4.1 leads with 68.0 among models released by April 2025

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep 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 68.0
  2. 2 GPT-4.1 mini 60.7
  3. 3 GPT-4o Mar 2025 58.8
  4. 4 Claude 3.7 Sonnet 58.4
  5. 5 Claude 3.5 Sonnet 56.5
  6. 6 Qwen2.5-VL-72B 54.5
  7. 7 Qwen2.5-VL-32B 54.1
  8. 8 Gemma 3 27B 50.5
  9. 9 InternVL3-78B 50.2
  10. 10 GPT-4o mini 50.0
  11. 11 InternVL3-38B 48.0
  12. 12 Llama 4 Scout 47.5
  13. 13 Gemma 3 12B 46.8
  14. 14 Qwen2.5-VL-7B 44.9
  15. 15 Claude 3 Haiku 44.8
  16. 16 InternVL3-14B 44.5
  17. 17 InternVL3-8B 41.2
  18. 18 Gemma 3 4B 40.5
  19. 19 Qwen2.5-VL-3B 38.0
  20. 20 LLaVA-OneVision 7B 37.8
  21. 21 InternVL3-2B 36.3
  22. 22 llava-v1.6-mistral-7b 35.4

Source: Official MIRAGE-MMST leaderboard

#ModelScoreSettingReleasedRun onSource
1 GPT-4.1 68.0 Apr 2025 Jun 2025 Official leaderboard
2 GPT-4.1 mini 60.7 Apr 2025 Jun 2025 Official leaderboard
3 GPT-4o Mar 2025 58.8 Mar 2025 Jun 2025 Official leaderboard
4 Claude 3.7 Sonnet 58.4 Feb 2025 Jun 2025 Official leaderboard
5 Claude 3.5 Sonnet 56.5 Jun 2024 Jun 2025 Official leaderboard
6 Qwen2.5-VL-72B 54.5 Jan 2025 Jun 2025 Official leaderboard
7 Qwen2.5-VL-32B 54.1 Mar 2025 Jun 2025 Official leaderboard
8 Gemma 3 27B 50.5 Mar 2025 Jun 2025 Official leaderboard
9 InternVL3-78B 50.2 Apr 2025 Jun 2025 Official leaderboard
10 GPT-4o mini 50.0 Jul 2024 Jun 2025 Official leaderboard
11 InternVL3-38B 48.0 Apr 2025 Jun 2025 Official leaderboard
12 Llama 4 Scout 47.5 Apr 2025 Jun 2025 Official leaderboard
13 Gemma 3 12B 46.8 Mar 2025 Jun 2025 Official leaderboard
14 Qwen2.5-VL-7B 44.9 Jan 2025 Jun 2025 Official leaderboard
15 Claude 3 Haiku 44.8 Mar 2024 Jun 2025 Official leaderboard
16 InternVL3-14B 44.5 Apr 2025 Jun 2025 Official leaderboard
17 InternVL3-8B 41.2 Apr 2025 Jun 2025 Official leaderboard
18 Gemma 3 4B 40.5 Mar 2025 Jun 2025 Official leaderboard
19 Qwen2.5-VL-3B 38.0 Jan 2025 Jun 2025 Official leaderboard
20 LLaVA-OneVision 7B 37.8 Aug 2024 Jun 2025 Official leaderboard
21 InternVL3-2B 36.3 Apr 2025 Jun 2025 Official leaderboard
22 llava-v1.6-mistral-7b 35.4 Jan 2024 Jun 2025 Official leaderboard

Rankings by part

We read 8 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as Identification weighted score, Management weighted score and Identification accuracy. GPT-4.1 leads every one of them. Arrows show how far a model moved from its overall rank among the models we track.

Identification weighted score

22 models we track have a score here.

  1. 1 GPT-4.1 54.4
  2. 2 GPT-4o Mar 2025 ↑1 47.0
  3. 3 GPT-4.1 mini ↓1 46.0
  4. 4 Claude 3.7 Sonnet 44.6
  5. 5 Claude 3.5 Sonnet 42.2
  6. 6 Qwen2.5-VL-72B 40.5
  7. 7 Qwen2.5-VL-32B 37.0
  8. 8 InternVL3-78B ↑1 33.6
  9. 9 GPT-4o mini ↑1 33.1
  10. 10 Gemma 3 27B ↓2 31.1

Number one over time

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

OpenAI
Own leaderboard Not tested since Jun 2025
GPT-4.1 68.0

Source: Official MIRAGE-MMST 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 2024 2025 30.0 40.0 50.0 60.0 70.0 2024 2025

Source: Official MIRAGE-MMST 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 Vision

All 636 in Vision
  • MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy. Qwen 3.6 Plus 86%
  • MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
  • CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
  • MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy. Seed2.1 Pro 90.7%
  • MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy. Qwen 3.8 Max 97.7%
  • MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy. Qwen 3.8 Max 85.9%

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 GPT-4.1. 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 (22 entries).
  • Pass Models scored 22 models we track.
  • Warning Newest score The newest came 19 months ago, in April 2025, for GPT-4.1.
  • 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 mirage-benchmark.github.io: 22 rows for models we track 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