Benchmark Almanac

AgroMind (original)

GPT-4o (snapshot unspecified) 43% tops its leaderboard InternVL2-26B 35.8% leads the models we track
Not in model cards

Models answer questions about farmland images across 13 tasks. They identify crops, assess crop health and interpret spatial relationships.

The original evaluation contains 25,026 questions about 15,556 images.

Made by
Qingmei Li and colleagues
Released
May 2025
Kind
General visual questions
Level
University
Scored
Checked automatically
Answers
Multiple choice
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Images → Text
Interaction
Single turn
Run by
Anyone
In views
Vision
Last model added
December 2024
Scores refreshed
November 20th, 2026
21
Entries in its paper, 15 models we track

GPT-4o (snapshot unspecified) tops the leaderboard with 43% InternVL2-26B leads with 35.8% among models released by December 2024

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 InternVL2-26B, at 35.8%. This view shows only models we track, each at its best.

The newest model with a score here came out in December 2024, 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 (snapshot unspecified) 43%
  2. 2 Gemini-1.5-Flash (snapshot unspecified) 41.2%
  3. 3 InternVL2-26B 35.8%
  4. 4 LLaVA-v1.6-34B 34.9%
  5. 5 InternVL2-8B 34.9%
  6. 6 Claude-3.5-Sonnet (snapshot unspecified) 33.3%
  7. 7 InternLM-XComposer2-4KHD-7B 32.3%
  8. 8 llama3-llava-next-8b 32.1%
  9. 9 Mantis-Idefics2 31.8%
  10. 10 Mantis-LLaMA3-SigLIP 31.6%
  11. 11 InternVL2-4B 31.5%
  12. 12 llava-v1.6-vicuna-13b 31.4%
  13. 13 llava-v1.6-mistral-7b 31.1%
  14. 14 Idefics2-8B 30.9%
  15. 15 TinyLLaVA 29.9%
  16. 16 InternVL2-2B 28.5%
  17. 17 DeepSeek-VL2-Small 27.8%
  18. 18 DeepSeek-VL2-Tiny 27.5%
  19. 19 llava-v1.6-vicuna-7b 24.6%
  20. 20 LLaVA-NeXT-Interleave-7B 17.8%
  21. 21 instructblip-vicuna-7b 17.7%

Source: The AgroMind (original) paper

#EntryScoreSettingDateSource
1 GPT-4o (snapshot unspecified) 43% Official leaderboard
2 Gemini-1.5-Flash (snapshot unspecified) 41.2% Official leaderboard
3 InternVL2-26B 35.8% Jul 2024 Official leaderboard
4 LLaVA-v1.6-34B 34.9% Jan 2024 Official leaderboard
5 InternVL2-8B 34.9% Jul 2024 Official leaderboard
6 Claude-3.5-Sonnet (snapshot unspecified) 33.3% Official leaderboard
7 InternLM-XComposer2-4KHD-7B 32.3% Apr 2024 Official leaderboard
8 llama3-llava-next-8b 32.1% Apr 2024 Official leaderboard
9 Mantis-Idefics2 8B 31.8% Official leaderboard
10 Mantis-LLaMA3-SigLIP 8B 31.6% Official leaderboard
11 InternVL2-4B 31.5% Jul 2024 Official leaderboard
12 llava-v1.6-vicuna-13b 31.4% Jan 2024 Official leaderboard
13 llava-v1.6-mistral-7b 31.1% Jan 2024 Official leaderboard
14 Idefics2-8B 30.9% Apr 2024 Official leaderboard
15 TinyLLaVA 3.1B 29.9% Official leaderboard
16 InternVL2-2B 28.5% Jul 2024 Official leaderboard
17 DeepSeek-VL2-Small 27.8% Dec 2024 Official leaderboard
18 DeepSeek-VL2-Tiny 27.5% Dec 2024 Official leaderboard
19 llava-v1.6-vicuna-7b 24.6% Jan 2024 Official leaderboard
20 LLaVA-NeXT-Interleave-7B 17.8% Jun 2024 Official leaderboard
21 instructblip-vicuna-7b 17.7% May 2023 Official leaderboard
#ModelScoreSettingDateSource
1 InternVL2-26B 35.8% Jul 2024 Official leaderboard
2 LLaVA-v1.6-34B 34.9% Jan 2024 Official leaderboard
3 InternVL2-8B 34.9% Jul 2024 Official leaderboard
4 InternLM-XComposer2-4KHD-7B 32.3% Apr 2024 Official leaderboard
5 llama3-llava-next-8b 32.1% Apr 2024 Official leaderboard
6 InternVL2-4B 31.5% Jul 2024 Official leaderboard
7 llava-v1.6-vicuna-13b 31.4% Jan 2024 Official leaderboard
8 llava-v1.6-mistral-7b 31.1% Jan 2024 Official leaderboard
9 Idefics2-8B 30.9% Apr 2024 Official leaderboard
10 InternVL2-2B 28.5% Jul 2024 Official leaderboard
11 DeepSeek-VL2-Small 27.8% Dec 2024 Official leaderboard
12 DeepSeek-VL2-Tiny 27.5% Dec 2024 Official leaderboard
13 llava-v1.6-vicuna-7b 24.6% Jan 2024 Official leaderboard
14 LLaVA-NeXT-Interleave-7B 17.8% Jun 2024 Official leaderboard
15 instructblip-vicuna-7b 17.7% May 2023 Official leaderboard

Rankings by part

We read 30 parts from the benchmark’s paper, each ranking models on one part of the test, such as SL, SR and BD. InternVL2-26B leads 8 of them, and 10 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

SL

15 models we track have a score here.

  1. 1 LLaVA-v1.6-34B ↑1 48.1%
  2. 2 llama3-llava-next-8b ↑3 45.4%
  3. 3 llava-v1.6-mistral-7b ↑5 41.1%
  4. 4 InternVL2-2B ↑6 40%
  5. 5 InternVL2-8B ↓2 39.3%
  6. 6 llava-v1.6-vicuna-13b ↑1 38.6%
  7. 7 Idefics2-8B ↑2 32.6%
  8. 8 InternLM-XComposer2-4KHD-7B ↓4 32.1%
  9. 9 InternVL2-4B ↓3 28.6%
  10. 10 InternVL2-26B ↓9 21.5%

Number one over time

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

Shanghai AI Lab
Benchmark paper Not tested since May 2025

Source: The AgroMind (original) 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 Best so far
0% 20% 40% 60% 80% 100% 2023 2024 2025 Near the ceiling (85%) InternVL2-26B LLaVA-v1.6-34B instructblip-vicuna-7b 0% 20% 40% 60% 80% 100% 2023 2024 2025 Near the ceiling (85%) InternVL2-26B instructblip-vicuna-7b

Source: Official AgroMind (original) 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

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 23 months ago, in December 2024, for DeepSeek-VL2-Small. 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 benchmark’s paper (21 entries).
  • Pass Models scored 15 models we track, plus 6 other entries on its leaderboard.
  • Warning Newest score The newest came 23 months ago, in December 2024, for DeepSeek-VL2-Small.
  • Note Frontier models None of its models are on the capabilities index.
  • Pass Headroom The best is 35.8%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 15 rows for models we track and 6 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