AgroMind (original)
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
- Interaction
- Single turn
- Items from
- Another benchmark, Generated
- Refresh
- New versions replace it
- Run by
- Anyone
- In views
- Vision
- Last model added
- December 2024
- Scores refreshed
- November 20th, 2026
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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 43.0 GPT-4o (snapshot unspecified)
- 41.2 Gemini-1.5-Flash (snapshot unspecified)
- 35.8
InternVL2-26B
- 34.9
LLaVA-v1.6-34B
- 34.9
InternVL2-8B
- 33.3 Claude-3.5-Sonnet (snapshot unspecified)
- 32.3
InternLM-XComposer2-4KHD-7B
- 32.1 llama3-llava-next-8b
- 31.8 Mantis-Idefics2
- 31.6 Mantis-LLaMA3-SigLIP
- 31.5
InternVL2-4B
- 31.4 llava-v1.6-vicuna-13b
- 31.1 llava-v1.6-mistral-7b
- 30.9
Idefics2-8B
- 29.9 TinyLLaVA
- 28.5
InternVL2-2B
- 27.8
DeepSeek-VL2-Small
- 27.5
DeepSeek-VL2-Tiny
- 24.6 llava-v1.6-vicuna-7b
- 17.8
LLaVA-NeXT-Interleave-7B
- 17.7 instructblip-vicuna-7b
- 1 GPT-4o (snapshot unspecified) 43%
- 2 Gemini-1.5-Flash (snapshot unspecified) 41.2%
- 3
InternVL2-26B 35.8%
- 4
LLaVA-v1.6-34B 34.9%
- 5
InternVL2-8B 34.9%
- 6 Claude-3.5-Sonnet (snapshot unspecified) 33.3%
- 7
InternLM-XComposer2-4KHD-7B 32.3%
- 8 llama3-llava-next-8b 32.1%
- 9 Mantis-Idefics2 31.8%
- 10 Mantis-LLaMA3-SigLIP 31.6%
- 11
InternVL2-4B 31.5%
- 12 llava-v1.6-vicuna-13b 31.4%
- 13 llava-v1.6-mistral-7b 31.1%
- 14
Idefics2-8B 30.9%
- 15 TinyLLaVA 29.9%
- 16
InternVL2-2B 28.5%
- 17
DeepSeek-VL2-Small 27.8%
- 18
DeepSeek-VL2-Tiny 27.5%
- 19 llava-v1.6-vicuna-7b 24.6%
- 20
LLaVA-NeXT-Interleave-7B 17.8%
- 21 instructblip-vicuna-7b 17.7%
Source: The AgroMind (original) paper
- 35.8
InternVL2-26B
- 34.9
LLaVA-v1.6-34B
- 34.9
InternVL2-8B
- 32.3
InternLM-XComposer2-4KHD-7B
- 32.1 llama3-llava-next-8b
- 31.5
InternVL2-4B
- 31.4 llava-v1.6-vicuna-13b
- 31.1 llava-v1.6-mistral-7b
- 30.9
Idefics2-8B
- 28.5
InternVL2-2B
- 27.8
DeepSeek-VL2-Small
- 27.5
DeepSeek-VL2-Tiny
- 24.6 llava-v1.6-vicuna-7b
- 17.8
LLaVA-NeXT-Interleave-7B
- 17.7 instructblip-vicuna-7b
- 1
InternVL2-26B 35.8%
- 2
LLaVA-v1.6-34B 34.9%
- 3
InternVL2-8B 34.9%
- 4
InternLM-XComposer2-4KHD-7B 32.3%
- 5 llama3-llava-next-8b 32.1%
- 6
InternVL2-4B 31.5%
- 7 llava-v1.6-vicuna-13b 31.4%
- 8 llava-v1.6-mistral-7b 31.1%
- 9
Idefics2-8B 30.9%
- 10
InternVL2-2B 28.5%
- 11
DeepSeek-VL2-Small 27.8%
- 12
DeepSeek-VL2-Tiny 27.5%
- 13 llava-v1.6-vicuna-7b 24.6%
- 14
LLaVA-NeXT-Interleave-7B 17.8%
- 15 instructblip-vicuna-7b 17.7%
Source: The AgroMind (original) paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-4o (snapshot unspecified) | 43% | Official leaderboard | ||
| 2 | Gemini-1.5-Flash (snapshot unspecified) | 41.2% | Official leaderboard | ||
| 3 | | 35.8% | Jul 2024 | Official leaderboard | |
| 4 | | 34.9% | Jan 2024 | Official leaderboard | |
| 5 | | 34.9% | Jul 2024 | Official leaderboard | |
| 6 | Claude-3.5-Sonnet (snapshot unspecified) | 33.3% | Official leaderboard | ||
| 7 | | 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 | | 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 | | 30.9% | Apr 2024 | Official leaderboard | |
| 15 | TinyLLaVA 3.1B | 29.9% | Official leaderboard | ||
| 16 | | 28.5% | Jul 2024 | Official leaderboard | |
| 17 | | 27.8% | Dec 2024 | Official leaderboard | |
| 18 | | 27.5% | Dec 2024 | Official leaderboard | |
| 19 | llava-v1.6-vicuna-7b | 24.6% | Jan 2024 | Official leaderboard | |
| 20 | | 17.8% | Jun 2024 | Official leaderboard | |
| 21 | instructblip-vicuna-7b | 17.7% | May 2023 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 35.8% | Jul 2024 | Official leaderboard | |
| 2 | | 34.9% | Jan 2024 | Official leaderboard | |
| 3 | | 34.9% | Jul 2024 | Official leaderboard | |
| 4 | | 32.3% | Apr 2024 | Official leaderboard | |
| 5 | llama3-llava-next-8b | 32.1% | Apr 2024 | Official leaderboard | |
| 6 | | 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 | | 30.9% | Apr 2024 | Official leaderboard | |
| 10 | | 28.5% | Jul 2024 | Official leaderboard | |
| 11 | | 27.8% | Dec 2024 | Official leaderboard | |
| 12 | | 27.5% | Dec 2024 | Official leaderboard | |
| 13 | llava-v1.6-vicuna-7b | 24.6% | Jan 2024 | Official leaderboard | |
| 14 | | 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
LLaVA-v1.6-34B ↑1 48.1%
- 2 llama3-llava-next-8b ↑3 45.4%
- 3 llava-v1.6-mistral-7b ↑5 41.1%
- 4
InternVL2-2B ↑6 40%
- 5
InternVL2-8B ↓2 39.3%
- 6 llava-v1.6-vicuna-13b ↑1 38.6%
- 7
Idefics2-8B ↑2 32.6%
- 8
InternLM-XComposer2-4KHD-7B ↓4 32.1%
- 9
InternVL2-4B ↓3 28.6%
- 10
InternVL2-26B ↓9 21.5%
SR
15 models we track have a score here.
- 1
InternVL2-2B ↑9 15%
- 2
Idefics2-8B ↑7 13.8%
- 3
InternVL2-26B ↓2 13.5%
- 4
DeepSeek-VL2-Small ↑7 13.3%
- 5
InternVL2-4B ↑1 12.5%
- 6 llama3-llava-next-8b ↓1 12.2%
- 7 llava-v1.6-mistral-7b ↑1 12%
- 8 llava-v1.6-vicuna-13b ↓1 12%
- 9 llava-v1.6-vicuna-7b ↑4 12%
- 10
DeepSeek-VL2-Tiny ↑2 12%
BD
15 models we track have a score here.
- 1
LLaVA-v1.6-34B ↑1 54.5%
- 2 llava-v1.6-mistral-7b ↑6 49.3%
- 3 llava-v1.6-vicuna-7b ↑10 48.8%
- 4 llava-v1.6-vicuna-13b ↑3 47.3%
- 5 llama3-llava-next-8b 44.5%
- 6
InternLM-XComposer2-4KHD-7B ↓2 40%
- 7
Idefics2-8B ↑2 33.5%
- 8
InternVL2-4B ↓2 28.8%
- 9
LLaVA-NeXT-Interleave-7B ↑5 25.3%
- 10
InternVL2-2B 24.8%
OC
15 models we track have a score here.
- 1
InternVL2-26B 69.5%
- 2
Idefics2-8B ↑7 60.3%
- 3
DeepSeek-VL2-Small ↑8 59.9%
- 4
DeepSeek-VL2-Tiny ↑8 57.9%
- 5
LLaVA-v1.6-34B ↓3 57.7%
- 6 llava-v1.6-vicuna-13b ↑1 55.1%
- 7
InternVL2-8B ↓4 52.2%
- 8 llama3-llava-next-8b ↓3 52%
- 9
InternVL2-4B ↓3 48.2%
- 10 llava-v1.6-vicuna-7b ↑3 47.7%
PDD
15 models we track have a score here.
- 1
InternVL2-4B ↑5 45.1%
- 2
InternVL2-26B ↓1 43.9%
- 3
LLaVA-v1.6-34B ↓1 43%
- 4 llava-v1.6-mistral-7b ↑4 42%
- 5
InternVL2-8B ↓2 41.9%
- 6
InternLM-XComposer2-4KHD-7B ↓2 37.2%
- 7
Idefics2-8B ↑2 36.3%
- 8 llama3-llava-next-8b ↓3 34.8%
- 9
DeepSeek-VL2-Small ↑2 31.1%
- 10 llava-v1.6-vicuna-7b ↑3 29.8%
GSR
15 models we track have a score here.
- 1 llava-v1.6-mistral-7b ↑7 61%
- 2
InternVL2-2B ↑8 47.5%
- 3
DeepSeek-VL2-Tiny ↑9 45.9%
- 4
Idefics2-8B ↑5 42.6%
- 5 llama3-llava-next-8b 39.9%
- 6
InternVL2-8B ↓3 38.3%
- 7
DeepSeek-VL2-Small ↑4 37.9%
- 8
LLaVA-v1.6-34B ↓6 37.1%
- 9
InternLM-XComposer2-4KHD-7B ↓5 35.8%
- 10 llava-v1.6-vicuna-13b ↓3 34.7%
SC
15 models we track have a score here.
- 1
InternVL2-8B ↑2 33.2%
- 2
InternLM-XComposer2-4KHD-7B ↑2 32.7%
- 3
InternVL2-26B ↓2 32.5%
- 4
InternVL2-4B ↑2 30.5%
- 5
DeepSeek-VL2-Small ↑6 28.4%
- 6
DeepSeek-VL2-Tiny ↑6 27.9%
- 7
InternVL2-2B ↑3 25.6%
- 8 llama3-llava-next-8b ↓3 25.6%
- 9
Idefics2-8B 23.7%
- 10
LLaVA-v1.6-34B ↓8 22.5%
CO
15 models we track have a score here.
- 1 llava-v1.6-vicuna-13b ↑6 29.9%
- 2 llama3-llava-next-8b ↑3 29.6%
- 3
LLaVA-v1.6-34B ↓1 29.1%
- 4 llava-v1.6-mistral-7b ↑4 28.6%
- 5
InternVL2-8B ↓2 28.5%
- 6
InternVL2-26B ↓5 28.4%
- 7 llava-v1.6-vicuna-7b ↑6 27.8%
- 8
InternLM-XComposer2-4KHD-7B ↓4 27.7%
- 9
DeepSeek-VL2-Small ↑2 27.7%
- 10
Idefics2-8B ↓1 27.6%
AS
15 models we track have a score here.
- 1
InternVL2-26B 45.1%
- 2 llava-v1.6-mistral-7b ↑6 39.9%
- 3
LLaVA-v1.6-34B ↓1 39.8%
- 4
InternVL2-4B ↑2 39.6%
- 5
InternVL2-8B ↓2 39.5%
- 6 llava-v1.6-vicuna-13b ↑1 38.6%
- 7 llama3-llava-next-8b ↓2 35.1%
- 8
InternLM-XComposer2-4KHD-7B ↓4 27.1%
- 9
InternVL2-2B ↑1 25.7%
- 10
Idefics2-8B ↓1 23.6%
VPR
15 models we track have a score here.
- 1
InternVL2-8B ↑2 19%
- 2
InternVL2-26B ↓1 17.6%
- 3
InternVL2-4B ↑3 16%
- 4
DeepSeek-VL2-Small ↑7 14.3%
- 5
DeepSeek-VL2-Tiny ↑7 14%
- 6
LLaVA-v1.6-34B ↓4 13.6%
- 7
InternVL2-2B ↑3 12.8%
- 8
LLaVA-NeXT-Interleave-7B ↑6 10.8%
- 9
InternLM-XComposer2-4KHD-7B ↓5 10.2%
- 10 llava-v1.6-vicuna-7b ↑3 9.91%
AR
15 models we track have a score here.
- 1
InternLM-XComposer2-4KHD-7B ↑3 38.1%
- 2
InternVL2-2B ↑8 28.8%
- 3
InternVL2-4B ↑3 24.1%
- 4 instructblip-vicuna-7b ↑11 17.2%
- 5
InternVL2-8B ↓2 16.8%
- 6
DeepSeek-VL2-Small ↑5 16.4%
- 7
Idefics2-8B ↑2 14.9%
- 8
LLaVA-NeXT-Interleave-7B ↑6 13.5%
- 9 llama3-llava-next-8b ↓4 13.1%
- 10
LLaVA-v1.6-34B ↓8 12.7%
CTR
15 models we track have a score here.
- 1
InternVL2-26B 57.6%
- 2
InternVL2-8B ↑1 48.3%
- 3
Idefics2-8B ↑6 44.1%
- 4
InternLM-XComposer2-4KHD-7B 43.5%
- 5
LLaVA-v1.6-34B ↓3 42.1%
- 6
InternVL2-4B 41%
- 7 llama3-llava-next-8b ↓2 37.6%
- 8
DeepSeek-VL2-Tiny ↑4 37%
- 9
InternVL2-2B ↑1 36.9%
- 10 llava-v1.6-vicuna-13b ↓3 32.5%
PL
15 models we track have a score here.
- 1
InternLM-XComposer2-4KHD-7B ↑3 38.5%
- 2 llava-v1.6-vicuna-13b ↑5 31.8%
- 3
InternVL2-8B 31.1%
- 4
LLaVA-v1.6-34B ↓2 27.6%
- 5
DeepSeek-VL2-Tiny ↑7 27%
- 6
InternVL2-26B ↓5 25.8%
- 7
InternVL2-4B ↓1 25.5%
- 8 llama3-llava-next-8b ↓3 25.3%
- 9
Idefics2-8B 23.7%
- 10 llava-v1.6-vicuna-7b ↑3 20.6%
Easy
15 models we track have a score here.
- 1
DeepSeek-VL2-Tiny ↑11 73.5%
- 2 llava-v1.6-mistral-7b ↑6 69.3%
- 3
InternVL2-26B ↓2 64.5%
- 4
InternVL2-8B ↓1 62.6%
- 5
InternVL2-2B ↑5 62.5%
- 6
Idefics2-8B ↑3 61.5%
- 7
LLaVA-v1.6-34B ↓5 61.3%
- 8
DeepSeek-VL2-Small ↑3 61%
- 9 llama3-llava-next-8b ↓4 60.4%
- 10 llava-v1.6-vicuna-13b ↓3 59.9%
Medium
15 models we track have a score here.
- 1
InternVL2-26B 50.7%
- 2
InternVL2-8B ↑1 45.7%
- 3
InternVL2-4B ↑3 44.8%
- 4
LLaVA-v1.6-34B ↓2 44.7%
- 5 llava-v1.6-vicuna-13b ↑2 42.3%
- 6
InternLM-XComposer2-4KHD-7B ↓2 41.6%
- 7 llama3-llava-next-8b ↓2 40.6%
- 8
Idefics2-8B ↑1 39.4%
- 9 llava-v1.6-mistral-7b ↓1 37.7%
- 10
DeepSeek-VL2-Small ↑1 35.2%
Hard
15 models we track have a score here.
- 1
LLaVA-v1.6-34B ↑1 20.8%
- 2
InternVL2-8B ↑1 19.5%
- 3 llama3-llava-next-8b ↑2 18.7%
- 4
InternVL2-2B ↑6 18.6%
- 5
InternLM-XComposer2-4KHD-7B ↓1 18.4%
- 6 llava-v1.6-mistral-7b ↑2 17.5%
- 7
Idefics2-8B ↑2 17%
- 8
InternVL2-26B ↓7 16.7%
- 9 llava-v1.6-vicuna-13b ↓2 15.9%
- 10
InternVL2-4B ↓4 15%
Judgment
15 models we track have a score here.
- 1
InternVL2-8B ↑2 71.1%
- 2
DeepSeek-VL2-Small ↑9 70.2%
- 3
DeepSeek-VL2-Tiny ↑9 69.8%
- 4 llama3-llava-next-8b ↑1 69.7%
- 5
Idefics2-8B ↑4 65.1%
- 6
InternLM-XComposer2-4KHD-7B ↓2 63.5%
- 7
InternVL2-26B ↓6 61.5%
- 8
InternVL2-2B ↑2 61.2%
- 9 llava-v1.6-vicuna-13b ↓2 60.6%
- 10
InternVL2-4B ↓4 59.6%
Counting
15 models we track have a score here.
- 1
DeepSeek-VL2-Small ↑10 32.5%
- 2
LLaVA-v1.6-34B 31%
- 3
InternVL2-26B ↓2 30.1%
- 4
InternVL2-8B ↓1 29.9%
- 5 llama3-llava-next-8b 28.4%
- 6 llava-v1.6-vicuna-7b ↑7 28.1%
- 7
Idefics2-8B ↑2 28%
- 8 llava-v1.6-vicuna-13b ↓1 27.9%
- 9 llava-v1.6-mistral-7b ↓1 27.6%
- 10
InternVL2-2B 26.8%
Text Choice
15 models we track have a score here.
- 1
InternVL2-26B 42.5%
- 2
LLaVA-v1.6-34B 41.9%
- 3
InternVL2-8B 39.9%
- 4 llava-v1.6-vicuna-13b ↑3 37%
- 5
InternVL2-4B ↑1 36.9%
- 6 llama3-llava-next-8b ↓1 36.7%
- 7 llava-v1.6-mistral-7b ↑1 36.4%
- 8
InternLM-XComposer2-4KHD-7B ↓4 36.2%
- 9
Idefics2-8B 34.2%
- 10
InternVL2-2B 30.7%
Image Choice
15 models we track have a score here.
- 1
InternLM-XComposer2-4KHD-7B ↑3 30.9%
- 2
InternVL2-26B ↓1 29.9%
- 3
InternVL2-8B 29.5%
- 4
LLaVA-v1.6-34B ↓2 25.8%
- 5
InternVL2-4B ↑1 25.1%
- 6
LLaVA-NeXT-Interleave-7B ↑8 22.2%
- 7 llama3-llava-next-8b ↓2 21.8%
- 8
DeepSeek-VL2-Tiny ↑4 21.7%
- 9 instructblip-vicuna-7b ↑6 21.5%
- 10
DeepSeek-VL2-Small ↑1 21.4%
Open-ended
15 models we track have a score here.
- 1
DeepSeek-VL2-Tiny ↑11 15.7%
- 2 llava-v1.6-vicuna-7b ↑11 11.8%
- 3
InternVL2-2B ↑7 11.6%
- 4
Idefics2-8B ↑5 11.2%
- 5
InternLM-XComposer2-4KHD-7B ↓1 11.1%
- 6
LLaVA-NeXT-Interleave-7B ↑8 9.78%
- 7 llava-v1.6-mistral-7b ↑1 9.43%
- 8
LLaVA-v1.6-34B ↓6 6.87%
- 9
InternVL2-8B ↓6 6.25%
- 10 llama3-llava-next-8b ↓5 6.03%
Plot
15 models we track have a score here.
- 1
LLaVA-v1.6-34B ↑1 22%
- 2 llava-v1.6-mistral-7b ↑6 21.6%
- 3
Idefics2-8B ↑6 19.8%
- 4 llava-v1.6-vicuna-13b ↑3 19.3%
- 5 llama3-llava-next-8b 17.6%
- 6
DeepSeek-VL2-Tiny ↑6 17.3%
- 7
InternVL2-8B ↓4 16.3%
- 8
InternVL2-2B ↑2 14.1%
- 9
LLaVA-NeXT-Interleave-7B ↑5 13.5%
- 10
InternLM-XComposer2-4KHD-7B ↓6 13.5%
Disease
15 models we track have a score here.
- 1
InternVL2-26B 30.5%
- 2
InternVL2-8B ↑1 29.8%
- 3
InternVL2-4B ↑3 29.6%
- 4
LLaVA-v1.6-34B ↓2 29.3%
- 5 llava-v1.6-mistral-7b ↑3 27.2%
- 6
InternLM-XComposer2-4KHD-7B ↓2 24%
- 7 llama3-llava-next-8b ↓2 24%
- 8
Idefics2-8B ↑1 23.9%
- 9
DeepSeek-VL2-Small ↑2 20.4%
- 10 llava-v1.6-vicuna-7b ↑3 18.6%
Fruit
15 models we track have a score here.
- 1
Idefics2-8B ↑8 53.5%
- 2 llava-v1.6-vicuna-13b ↑5 53.3%
- 3
LLaVA-v1.6-34B ↓1 48.7%
- 4
DeepSeek-VL2-Small ↑7 47.5%
- 5 llama3-llava-next-8b 46.3%
- 6 llava-v1.6-mistral-7b ↑2 45.8%
- 7
InternVL2-26B ↓6 45.8%
- 8
InternVL2-4B ↓2 41.3%
- 9
InternVL2-2B ↑1 40.4%
- 10
InternVL2-8B ↓7 40.1%
Pest
15 models we track have a score here.
- 1
InternVL2-26B 63.5%
- 2
DeepSeek-VL2-Small ↑9 57.4%
- 3
DeepSeek-VL2-Tiny ↑9 56.5%
- 4
Idefics2-8B ↑5 55.6%
- 5
LLaVA-v1.6-34B ↓3 53%
- 6
InternVL2-8B ↓3 51.9%
- 7 llava-v1.6-vicuna-13b 49.9%
- 8 llama3-llava-next-8b ↓3 49.3%
- 9
InternLM-XComposer2-4KHD-7B ↓5 47.9%
- 10
InternVL2-4B ↓4 47.5%
Tassel
15 models we track have a score here.
- 1
InternVL2-8B ↑2 27.8%
- 2
DeepSeek-VL2-Tiny ↑10 27.1%
- 3
DeepSeek-VL2-Small ↑8 26.1%
- 4
InternLM-XComposer2-4KHD-7B 24.2%
- 5
InternVL2-4B ↑1 22.9%
- 6
InternVL2-2B ↑4 22.5%
- 7 llava-v1.6-vicuna-7b ↑6 22.1%
- 8
InternVL2-26B ↓7 22%
- 9 llama3-llava-next-8b ↓4 19.1%
- 10
LLaVA-NeXT-Interleave-7B ↑4 18.5%
Crop
15 models we track have a score here.
- 1
InternLM-XComposer2-4KHD-7B ↑3 44.1%
- 2 llava-v1.6-vicuna-13b ↑5 34.6%
- 3
LLaVA-v1.6-34B ↓1 33.8%
- 4
InternVL2-8B ↓1 33.7%
- 5
InternVL2-26B ↓4 32.1%
- 6
InternVL2-4B 30.8%
- 7 llama3-llava-next-8b ↓2 25.7%
- 8
LLaVA-NeXT-Interleave-7B ↑6 24.1%
- 9
InternVL2-2B ↑1 18.9%
- 10 llava-v1.6-vicuna-7b ↑3 18.3%
Anomaly
15 models we track have a score here.
- 1 llava-v1.6-mistral-7b ↑7 38%
- 2
LLaVA-v1.6-34B 37.6%
- 3 llama3-llava-next-8b ↑2 37.5%
- 4
InternVL2-8B ↓1 35.5%
- 5
InternVL2-26B ↓4 34.4%
- 6
InternVL2-4B 33.1%
- 7
InternLM-XComposer2-4KHD-7B ↓3 32.4%
- 8 llava-v1.6-vicuna-13b ↓1 31.2%
- 9
InternVL2-2B ↑1 30.9%
- 10
Idefics2-8B ↓1 22.4%
Oil Palm
15 models we track have a score here.
- 1
DeepSeek-VL2-Tiny ↑11 36.8%
- 2 llava-v1.6-mistral-7b ↑6 34.4%
- 3
InternVL2-2B ↑7 29.4%
- 4
InternLM-XComposer2-4KHD-7B 25.6%
- 5
InternVL2-8B ↓2 25.4%
- 6
Idefics2-8B ↑3 23.5%
- 7
DeepSeek-VL2-Small ↑4 18.7%
- 8 instructblip-vicuna-7b ↑7 18.6%
- 9 llava-v1.6-vicuna-7b ↑4 18%
- 10
InternVL2-26B ↓9 17.4%
Tree Cover
15 models we track have a score here.
- 1
InternVL2-26B 52.2%
- 2
InternVL2-8B ↑1 47.9%
- 3
LLaVA-v1.6-34B ↓1 46%
- 4
InternVL2-4B ↑2 42.1%
- 5
InternLM-XComposer2-4KHD-7B ↓1 41.7%
- 6 llama3-llava-next-8b ↓1 41%
- 7 llava-v1.6-vicuna-13b 38.5%
- 8
Idefics2-8B ↑1 37.3%
- 9 llava-v1.6-mistral-7b ↓1 36%
- 10
DeepSeek-VL2-Small ↑1 33.4%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
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.
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
- 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.