KORGym
Models play 51 text and visual games designed to test interactive reasoning.
The May 2025 paper compares models across 51 games. The abstract does not state the best score.
- Made by
- KORGym authors
- Released
- May 2025
- Kind
- Puzzles and chess
- Level
- Everyday
- Scored
- Settled by events
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Agentic
- Environment
- KORGym
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
-
Gemini 2.5 Pro Mar 2025 #104 on the capability index
- Last model added
- April 2025
- Scores refreshed
- November 20th, 2026
o3-mini leads with 0.82 among models released by April 2025 o3-mini leads with 0.82 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:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 0.82 o3-mini
- 0.79
Gemini 2.5 Pro
- 0.73 o1
- 0.72 Doubao-1-5-thinking-pro
- 0.71 DeepSeek-R1 (snapshot unspecified)
- 0.62 Claude 3.7 Sonnet
- 0.60
Qwen3-32B
- 0.42
DeepSeek-V3
- 0.39
DeepSeek-R1-Distill-Qwen-32B
- 0.38
Gemini 2.0 Flash Thinking
- 0.28 Qwen-QwQ
- 0.26
Gemini 2.0 Flash
- 0.22 GPT-4o (snapshot unspecified)
- 0.20
Doubao-1.5-pro
- 0.18
Qwen2.5-72B
- 0.16 Qwen2.5-32B-Instruct
- 0.11
DeepSeek-R1-Distill-Qwen-7B
- 0.08 Qwen2.5-7B-Instruct
- 1 o3-mini Thinking 0.82
- 2
Gemini 2.5 Pro Mar 2025, Thinking 0.79
- 3 o1 Thinking 0.73
- 4 Doubao-1-5-thinking-pro Thinking 0.72
- 5 DeepSeek-R1 (snapshot unspecified) Thinking 0.71
- 6 Claude 3.7 Sonnet Thinking 0.62
- 7
Qwen3-32B Thinking 0.60
- 8
DeepSeek-V3 Mar 2025, Non-thinking 0.42
- 9
DeepSeek-R1-Distill-Qwen-32B Thinking 0.39
- 10
Gemini 2.0 Flash Thinking Jan 2025, Thinking 0.38
- 11 Qwen-QwQ Thinking 0.28
- 12
Gemini 2.0 Flash Feb 2025, Non-thinking 0.26
- 13 GPT-4o (snapshot unspecified) Non-thinking 0.22
- 14
Doubao-1.5-pro Non-thinking 0.20
- 15
Qwen2.5-72B Non-thinking 0.18
- 16 Qwen2.5-32B-Instruct Non-thinking 0.16
- 17
DeepSeek-R1-Distill-Qwen-7B Non-thinking 0.11
- 18 Qwen2.5-7B-Instruct Non-thinking 0.08
Source: Official KORGym leaderboard
- 1 o3-mini Thinking 0.82
- 2
Gemini 2.5 Pro Mar 2025, Thinking 0.79
- 3 o1 Thinking 0.73
- 4 Claude 3.7 Sonnet Thinking 0.62
- 5
Qwen3-32B Thinking 0.60
- 6
DeepSeek-V3 Mar 2025, Non-thinking 0.42
- 7
DeepSeek-R1-Distill-Qwen-32B Thinking 0.39
- 8
Gemini 2.0 Flash Thinking Jan 2025, Thinking 0.38
- 9
Gemini 2.0 Flash Feb 2025, Non-thinking 0.26
- 10
Doubao-1.5-pro Non-thinking 0.20
- 11
Qwen2.5-72B Non-thinking 0.18
- 12
DeepSeek-R1-Distill-Qwen-7B Non-thinking 0.11
Source: Official KORGym leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o3-mini | 0.82 | Thinking | Jan 2025 | Official leaderboard |
| 2 | | 0.79 | Thinking | Mar 2025 | Official leaderboard |
| 3 | o1 | 0.73 | Thinking | Dec 2024 | Official leaderboard |
| 4 | Doubao-1-5-thinking-pro | 0.72 | Thinking | Official leaderboard | |
| 5 | DeepSeek-R1 (snapshot unspecified) | 0.71 | Thinking | Official leaderboard | |
| 6 | Claude 3.7 Sonnet | 0.62 | Thinking | Feb 2025 | Official leaderboard |
| 7 | | 0.60 | Thinking | Apr 2025 | Official leaderboard |
| 8 | | 0.42 | Non-thinking | Mar 2025 | Official leaderboard |
| 9 | | 0.39 | Thinking | Jan 2025 | Official leaderboard |
| 10 | | 0.38 | Thinking | Jan 2025 | Official leaderboard |
| 11 | Qwen-QwQ | 0.28 | Thinking | Official leaderboard | |
| 12 | | 0.26 | Non-thinking | Feb 2025 | Official leaderboard |
| 13 | GPT-4o (snapshot unspecified) | 0.22 | Non-thinking | Official leaderboard | |
| 14 | | 0.20 | Non-thinking | Jan 2025 | Official leaderboard |
| 15 | | 0.18 | Non-thinking | Sep 2024 | Official leaderboard |
| 16 | Qwen2.5-32B-Instruct | 0.16 | Non-thinking | Official leaderboard | |
| 17 | | 0.11 | Non-thinking | Jan 2025 | Official leaderboard |
| 18 | Qwen2.5-7B-Instruct | 0.08 | Non-thinking | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o3-mini | 0.82 | Thinking | Jan 2025 | Official leaderboard |
| 2 | | 0.79 | Thinking | Mar 2025 | Official leaderboard |
| 3 | o1 | 0.73 | Thinking | Dec 2024 | Official leaderboard |
| 4 | Claude 3.7 Sonnet | 0.62 | Thinking | Feb 2025 | Official leaderboard |
| 5 | | 0.60 | Thinking | Apr 2025 | Official leaderboard |
| 6 | | 0.42 | Non-thinking | Mar 2025 | Official leaderboard |
| 7 | | 0.39 | Thinking | Jan 2025 | Official leaderboard |
| 8 | | 0.38 | Thinking | Jan 2025 | Official leaderboard |
| 9 | | 0.26 | Non-thinking | Feb 2025 | Official leaderboard |
| 10 | | 0.20 | Non-thinking | Jan 2025 | Official leaderboard |
| 11 | | 0.18 | Non-thinking | Sep 2024 | Official leaderboard |
| 12 | | 0.11 | Non-thinking | Jan 2025 | Official leaderboard |
Rankings by part
We read 47 parts from the benchmark’s own leaderboard in 5 groups: strategy, spatial, puzzle, math and control. Each ranks models on one part of the test. Gemini 2.5 Pro leads 15 of them, and 5 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Control and Interaction
12 models we track have a score here.
- 1 o3-mini 0.77
- 2 o1 ↑1 0.74
- 3
Gemini 2.5 Pro Mar 2025 ↓1 0.63
- 4
Qwen3-32B ↑1 0.58
- 5 Claude 3.7 Sonnet ↓1 0.50
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.45
- 7
DeepSeek-V3 Mar 2025 ↓1 0.35
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.25
- 9
Gemini 2.0 Flash Feb 2025 0.24
- 10
Qwen2.5-72B ↑1 0.18
Puzzle
12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.94
- 2 Claude 3.7 Sonnet ↑2 0.93
- 3 o1 0.83
- 4 o3-mini ↓3 0.81
- 5
DeepSeek-V3 Mar 2025 ↑1 0.55
- 6
Qwen3-32B ↓1 0.55
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.53
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 0.28
- 9
Gemini 2.0 Flash Feb 2025 0.28
- 10
Doubao-1.5-pro 0.16
Mathematical and Logical
12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.93
- 2 o3-mini ↓1 0.79
- 3 o1 0.65
- 4
Qwen3-32B ↑1 0.58
- 5 Claude 3.7 Sonnet ↓1 0.52
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.35
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.34
- 8
DeepSeek-V3 Mar 2025 ↓2 0.27
- 9
Gemini 2.0 Flash Feb 2025 0.17
- 10
Doubao-1.5-pro 0.16
Spatial and Geometric
12 models we track have a score here.
- 1 o3-mini 0.94
- 2 o1 ↑1 0.79
- 3
Gemini 2.5 Pro Mar 2025 ↓1 0.59
- 4
Qwen3-32B ↑1 0.55
- 5 Claude 3.7 Sonnet ↓1 0.53
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.33
- 7
DeepSeek-V3 Mar 2025 ↓1 0.26
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.18
- 9
Gemini 2.0 Flash Feb 2025 0.12
- 10
Qwen2.5-72B ↑1 0.07
Strategic
12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.84
- 2 o3-mini ↓1 0.76
- 3
Qwen3-32B ↑2 0.71
- 4
DeepSeek-V3 Mar 2025 ↑2 0.69
- 5 o1 ↓2 0.66
- 6 Claude 3.7 Sonnet ↓2 0.64
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.58
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 0.56
- 9
Gemini 2.0 Flash Feb 2025 0.51
- 10
Qwen2.5-72B ↑1 0.49
2048
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 2083.2
- 2 o1 ↑1 1652.2
- 3
Gemini 2.0 Flash Thinking Jan 2025 ↑5 1378.2
- 4
Qwen3-32B ↑1 1378.2
- 5 o3-mini ↓4 1285.6
- 6 Claude 3.7 Sonnet ↓2 1283.6
- 7
DeepSeek-V3 Mar 2025 ↓1 1101.4
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 990.0
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 884.2
- 10
Gemini 2.0 Flash Feb 2025 ↓1 879.8
Trust Evolution
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 66.5
- 2 o1 ↑1 64.5
- 3
Qwen3-32B ↑2 63.6
- 4
Gemini 2.0 Flash Feb 2025 ↑5 51.8
- 5
Qwen2.5-72B ↑6 47.1
- 6 Claude 3.7 Sonnet ↓2 38.0
- 7
DeepSeek-V3 Mar 2025 ↓1 37.8
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 37.7
- 9
Gemini 2.0 Flash Thinking Jan 2025 ↓1 37.5
- 10
DeepSeek-R1-Distill-Qwen-7B ↑2 36.4
N point
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 8.60
- 2 Claude 3.7 Sonnet ↑2 8.30
- 3
DeepSeek-V3 Mar 2025 ↑3 8.22
- 4
DeepSeek-R1-Distill-Qwen-32B ↑3 8.05
- 5
Qwen3-32B 7.95
- 6
Gemini 2.0 Flash Thinking Jan 2025 ↑2 7.92
- 7 o3-mini ↓6 7.83
- 8
Gemini 2.5 Pro Mar 2025 ↓6 7.75
- 9
Qwen2.5-72B ↑2 7.60
- 10
Doubao-1.5-pro 7.50
Spider Solitaire
Raw game score. 12 models we track have a score here.
- 1
Qwen2.5-72B ↑10 0.00
- 2 o1 ↑1 0.00
- 3
DeepSeek-R1-Distill-Qwen-32B ↑4 0.00
- 4
DeepSeek-R1-Distill-Qwen-7B ↑8 0.00
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.00
- 6
Doubao-1.5-pro ↑4 0.00
- 7 o3-mini ↓6 0.00
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.00
- 9 Claude 3.7 Sonnet ↓5 0.00
- 10
DeepSeek-V3 Mar 2025 ↓4 0.00
Circle the cat
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.30
- 2
Gemini 2.5 Pro Mar 2025 0.25
- 3
DeepSeek-V3 Mar 2025 ↑3 0.20
- 4
Qwen3-32B ↑1 0.15
- 5 Claude 3.7 Sonnet ↓1 0.10
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.05
- 7
Qwen2.5-72B ↑4 0.00
- 8 o1 ↓5 0.00
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 0.00
- 10
Gemini 2.0 Flash Thinking Jan 2025 ↓2 0.00
Maze
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.86
- 2
Qwen3-32B ↑3 0.74
- 3
Gemini 2.5 Pro Mar 2025 ↓1 0.50
- 4 Claude 3.7 Sonnet 0.40
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.38
- 6
DeepSeek-V3 Mar 2025 0.38
- 7 o1 ↓4 0.36
- 8
DeepSeek-R1-Distill-Qwen-7B ↑4 0.12
- 9
Qwen2.5-72B ↑2 0.04
- 10
Gemini 2.0 Flash Thinking Jan 2025 ↓2 0.02
Sokoban
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.78
- 2 o1 ↑1 0.70
- 3 Claude 3.7 Sonnet ↑1 0.26
- 4
Qwen3-32B ↑1 0.22
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.16
- 6
Gemini 2.5 Pro Mar 2025 ↓4 0.10
- 7
DeepSeek-V3 Mar 2025 ↓1 0.08
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.06
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 0.04
- 10
Qwen2.5-72B ↑1 0.02
play lines
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.84
- 2 o3-mini ↓1 0.72
- 3
Gemini 2.5 Pro Mar 2025 ↓1 0.66
- 4
Qwen3-32B ↑1 0.52
- 5 Claude 3.7 Sonnet ↓1 0.24
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.14
- 7
Doubao-1.5-pro ↑3 0.08
- 8
DeepSeek-V3 Mar 2025 ↓2 0.08
- 9
Qwen2.5-72B ↑2 0.02
- 10
DeepSeek-R1-Distill-Qwen-7B ↑2 0.00
Emoji Connect
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.96
- 2 o1 ↑1 0.94
- 3 Claude 3.7 Sonnet ↑1 0.76
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.62
- 5
DeepSeek-V3 Mar 2025 ↑1 0.56
- 6
Qwen3-32B ↓1 0.52
- 7
DeepSeek-R1-Distill-Qwen-32B 0.22
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.18
- 9
Gemini 2.0 Flash Feb 2025 0.12
- 10
Doubao-1.5-pro 0.08
8-puzzle
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.94
- 2 o1 ↑1 0.90
- 3
Gemini 2.5 Pro Mar 2025 ↓1 0.72
- 4
Qwen3-32B ↑1 0.56
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.48
- 6 Claude 3.7 Sonnet ↓2 0.46
- 7
DeepSeek-R1-Distill-Qwen-32B 0.38
- 8
DeepSeek-V3 Mar 2025 ↓2 0.22
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 0.10
- 10
Doubao-1.5-pro 0.08
Bubble Ball Sorting
Raw game score. 12 models we track have a score here.
- 1
Qwen3-32B ↑4 1.00
- 2 o1 ↑1 0.95
- 3 o3-mini ↓2 0.95
- 4 Claude 3.7 Sonnet 0.95
- 5
Gemini 2.5 Pro Mar 2025 ↓3 0.95
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.90
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.65
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.60
- 9
Qwen2.5-72B ↑2 0.35
- 10
Doubao-1.5-pro 0.30
Pipe Game
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.84
- 2
Gemini 2.5 Pro Mar 2025 0.74
- 3 o3-mini ↓2 0.54
- 4
Qwen3-32B ↑1 0.50
- 5 Claude 3.7 Sonnet ↓1 0.10
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.04
- 7
DeepSeek-V3 Mar 2025 ↓1 0.02
- 8
Qwen2.5-72B ↑3 0.00
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 0.00
- 10
Gemini 2.0 Flash Thinking Jan 2025 ↓2 0.00
Free the Key
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.85
- 2 o3-mini ↓1 0.85
- 3 Claude 3.7 Sonnet ↑1 0.85
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.85
- 5
Qwen3-32B 0.80
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.70
- 7
DeepSeek-V3 Mar 2025 ↓1 0.65
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.20
- 9
Qwen2.5-72B ↑2 0.10
- 10
Doubao-1.5-pro 0.05
Map Simulation
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.38
- 2 o1 ↑1 0.28
- 3 Claude 3.7 Sonnet ↑1 0.12
- 4
Gemini 2.0 Flash Feb 2025 ↑5 0.10
- 5
Gemini 2.5 Pro Mar 2025 ↓3 0.10
- 6
DeepSeek-V3 Mar 2025 0.06
- 7
Qwen3-32B ↓2 0.06
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 0.04
- 9
Gemini 2.0 Flash Thinking Jan 2025 ↓1 0.02
- 10
Doubao-1.5-pro 0.02
Arrow-pathway
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.30
- 2 Claude 3.7 Sonnet ↑2 0.18
- 3
Gemini 2.0 Flash Thinking Jan 2025 ↑5 0.02
- 4
Qwen2.5-72B ↑7 0.00
- 5 o1 ↓2 0.00
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.00
- 7
DeepSeek-R1-Distill-Qwen-7B ↑5 0.00
- 8
Doubao-1.5-pro ↑2 0.00
- 9
Gemini 2.0 Flash Feb 2025 0.00
- 10
DeepSeek-V3 Mar 2025 ↓4 0.00
Word Problem
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.96
- 2
Gemini 2.5 Pro Mar 2025 0.90
- 3 o3-mini ↓2 0.88
- 4 Claude 3.7 Sonnet 0.82
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.62
- 6
Qwen3-32B ↓1 0.48
- 7
DeepSeek-V3 Mar 2025 ↓1 0.46
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 0.34
- 9
Gemini 2.0 Flash Feb 2025 0.34
- 10
Doubao-1.5-pro 0.12
Alphabetical sorting
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 1.00
- 2 o3-mini ↓1 0.98
- 3 Claude 3.7 Sonnet ↑1 0.98
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.96
- 5
DeepSeek-V3 Mar 2025 ↑1 0.84
- 6
Gemini 2.0 Flash Thinking Jan 2025 ↑2 0.78
- 7
Gemini 2.0 Flash Feb 2025 ↑2 0.56
- 8
Qwen3-32B ↓3 0.56
- 9
DeepSeek-R1-Distill-Qwen-32B ↓2 0.50
- 10
Doubao-1.5-pro 0.36
Letter Connection
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.98
- 2 o3-mini ↓1 0.98
- 3 Claude 3.7 Sonnet ↑1 0.96
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.94
- 5
Qwen3-32B 0.68
- 6
DeepSeek-V3 Mar 2025 0.42
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.40
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.34
- 9
DeepSeek-R1-Distill-Qwen-32B ↓2 0.22
- 10
Qwen2.5-72B ↑1 0.16
Word Transformation
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.78
- 2 Claude 3.7 Sonnet ↑2 0.56
- 3 o1 0.48
- 4
Qwen3-32B ↑1 0.48
- 5 o3-mini ↓4 0.40
- 6
DeepSeek-V3 Mar 2025 0.38
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.18
- 8
DeepSeek-R1-Distill-Qwen-32B ↓1 0.14
- 9
Doubao-1.5-pro ↑1 0.14
- 10
Gemini 2.0 Flash Feb 2025 ↓1 0.12
Wordle
Raw game score. 12 models we track have a score here.
- 1 Claude 3.7 Sonnet ↑3 0.85
- 2
Gemini 2.5 Pro Mar 2025 0.65
- 3
Gemini 2.0 Flash Thinking Jan 2025 ↑5 0.55
- 4
DeepSeek-V3 Mar 2025 ↑2 0.50
- 5
Qwen3-32B 0.50
- 6 o1 ↓3 0.45
- 7
DeepSeek-R1-Distill-Qwen-32B 0.45
- 8 o3-mini ↓7 0.40
- 9
Gemini 2.0 Flash Feb 2025 0.25
- 10
Qwen2.5-72B ↑1 0.20
Crypto Word
Raw game score. 12 models we track have a score here.
- 1 o3-mini 1.00
- 2 Claude 3.7 Sonnet ↑2 1.00
- 3
Gemini 2.5 Pro Mar 2025 ↓1 1.00
- 4 o1 ↓1 0.85
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.50
- 6
DeepSeek-V3 Mar 2025 0.50
- 7
Qwen3-32B ↓2 0.45
- 8
Doubao-1.5-pro ↑2 0.15
- 9
Gemini 2.0 Flash Feb 2025 0.10
- 10
Qwen2.5-72B ↑1 0.00
Date calculation
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.26
- 2
Gemini 2.5 Pro Mar 2025 0.24
- 3 o3-mini ↓2 0.16
- 4
Gemini 2.0 Flash Thinking Jan 2025 ↑4 0.14
- 5
Qwen3-32B 0.14
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.12
- 7
DeepSeek-V3 Mar 2025 ↓1 0.12
- 8 Claude 3.7 Sonnet ↓4 0.08
- 9
DeepSeek-R1-Distill-Qwen-7B ↑3 0.06
- 10
Doubao-1.5-pro 0.06
Sudoku
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.64
- 2 o3-mini ↓1 0.42
- 3
Qwen3-32B ↑2 0.36
- 4
DeepSeek-R1-Distill-Qwen-32B ↑3 0.20
- 5
Gemini 2.0 Flash Feb 2025 ↑4 0.16
- 6 Claude 3.7 Sonnet ↓2 0.14
- 7
Doubao-1.5-pro ↑3 0.12
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.10
- 9 o1 ↓6 0.04
- 10
Qwen2.5-72B ↑1 0.02
Light out game
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 1.00
- 2 o3-mini ↓1 0.98
- 3
Qwen3-32B ↑2 0.96
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.80
- 5
DeepSeek-V3 Mar 2025 ↑1 0.30
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.26
- 7 Claude 3.7 Sonnet ↓3 0.16
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.12
- 9
Qwen2.5-72B ↑2 0.06
- 10
DeepSeek-R1-Distill-Qwen-7B ↑2 0.06
Square Addition
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.88
- 2
Gemini 2.5 Pro Mar 2025 0.76
- 3
Qwen3-32B ↑2 0.72
- 4 o1 ↓1 0.70
- 5 Claude 3.7 Sonnet ↓1 0.64
- 6
DeepSeek-V3 Mar 2025 0.46
- 7
DeepSeek-R1-Distill-Qwen-32B 0.22
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.14
- 9
Gemini 2.0 Flash Feb 2025 0.06
- 10
Qwen2.5-72B ↑1 0.00
Alien
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.84
- 2
Gemini 2.5 Pro Mar 2025 0.78
- 3 o3-mini ↓2 0.72
- 4 Claude 3.7 Sonnet 0.72
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.40
- 6
Qwen3-32B ↓1 0.40
- 7
DeepSeek-V3 Mar 2025 ↓1 0.26
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.22
- 9
Doubao-1.5-pro ↑1 0.20
- 10
Gemini 2.0 Flash Feb 2025 ↓1 0.16
Party Time
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.92
- 2 Claude 3.7 Sonnet ↑2 0.88
- 3 o3-mini ↓2 0.78
- 4 o1 ↓1 0.60
- 5
Qwen3-32B 0.60
- 6
Gemini 2.0 Flash Thinking Jan 2025 ↑2 0.56
- 7
DeepSeek-R1-Distill-Qwen-32B 0.54
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.28
- 9
Doubao-1.5-pro ↑1 0.24
- 10
DeepSeek-V3 Mar 2025 ↓4 0.14
Path Planning Problem
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.60
- 2
Gemini 2.5 Pro Mar 2025 0.54
- 3 Claude 3.7 Sonnet ↑1 0.48
- 4
Qwen3-32B ↑1 0.46
- 5 o1 ↓2 0.34
- 6
Gemini 2.0 Flash Feb 2025 ↑3 0.20
- 7
DeepSeek-V3 Mar 2025 ↓1 0.20
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.14
- 9
Doubao-1.5-pro ↑1 0.14
- 10
DeepSeek-R1-Distill-Qwen-32B ↓3 0.12
Construction Company
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.92
- 2 o3-mini ↓1 0.86
- 3 Claude 3.7 Sonnet ↑1 0.50
- 4
Qwen3-32B ↑1 0.30
- 5 o1 ↓2 0.28
- 6
DeepSeek-V3 Mar 2025 0.16
- 7
DeepSeek-R1-Distill-Qwen-32B 0.10
- 8
Qwen2.5-72B ↑3 0.06
- 9
Gemini 2.0 Flash Thinking Jan 2025 ↓1 0.04
- 10
Doubao-1.5-pro 0.02
One Stroke Drawing
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 1.00
- 2 o1 ↑1 0.76
- 3 o3-mini ↓2 0.76
- 4 Claude 3.7 Sonnet 0.68
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.32
- 6
Qwen3-32B ↓1 0.26
- 7
DeepSeek-V3 Mar 2025 ↓1 0.20
- 8
Doubao-1.5-pro ↑2 0.06
- 9
Qwen2.5-72B ↑2 0.02
- 10
DeepSeek-R1-Distill-Qwen-32B ↓3 0.02
Coloring issue
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.98
- 2 o3-mini ↓1 0.96
- 3
Qwen3-32B ↑2 0.94
- 4 o1 ↓1 0.80
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.74
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 0.70
- 7 Claude 3.7 Sonnet ↓3 0.70
- 8
DeepSeek-R1-Distill-Qwen-7B ↑4 0.14
- 9
DeepSeek-V3 Mar 2025 ↓3 0.14
- 10
Gemini 2.0 Flash Feb 2025 ↓1 0.08
City Traveller
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.84
- 2 Claude 3.7 Sonnet ↑2 0.70
- 3
Gemini 2.0 Flash Thinking Jan 2025 ↑5 0.50
- 4
Gemini 2.0 Flash Feb 2025 ↑5 0.44
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.38
- 6
Doubao-1.5-pro ↑4 0.34
- 7
DeepSeek-V3 Mar 2025 ↓1 0.32
- 8 o1 ↓5 0.30
- 9 o3-mini ↓8 0.28
- 10
Qwen3-32B ↓5 0.20
Minigrid
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.40
- 2 Claude 3.7 Sonnet ↑2 0.35
- 3 o3-mini ↓2 0.30
- 4
DeepSeek-V3 Mar 2025 ↑2 0.25
- 5 o1 ↓2 0.15
- 6
Gemini 2.0 Flash Feb 2025 ↑3 0.15
- 7
Qwen3-32B ↓2 0.15
- 8
DeepSeek-R1-Distill-Qwen-7B ↑4 0.12
- 9
DeepSeek-R1-Distill-Qwen-32B ↓2 0.10
- 10
Gemini 2.0 Flash Thinking Jan 2025 ↓2 0.10
Snake
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 17.5
- 2 o3-mini ↓1 15.9
- 3
Qwen3-32B ↑2 13.9
- 4
Gemini 2.5 Pro Mar 2025 ↓2 13.5
- 5 Claude 3.7 Sonnet ↓1 9.75
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 7.48
- 7
DeepSeek-V3 Mar 2025 ↓1 4.25
- 8
Gemini 2.0 Flash Feb 2025 ↑1 2.05
- 9
Gemini 2.0 Flash Thinking Jan 2025 ↓1 1.85
- 10
Doubao-1.5-pro 1.80
Tetris
Raw game score. 12 models we track have a score here.
- 1 o3-mini 0.90
- 2 o1 ↑1 0.55
- 3
DeepSeek-R1-Distill-Qwen-32B ↑4 0.30
- 4
Qwen3-32B ↑1 0.25
- 5 Claude 3.7 Sonnet ↓1 0.15
- 6
Gemini 2.0 Flash Thinking Jan 2025 ↑2 0.05
- 7
DeepSeek-V3 Mar 2025 ↓1 0.05
- 8
Gemini 2.5 Pro Mar 2025 ↓6 0.05
- 9
Qwen2.5-72B ↑2 0.00
- 10
DeepSeek-R1-Distill-Qwen-7B ↑2 0.00
Tower of Hanoi
Raw game score. 12 models we track have a score here.
- 1 o3-mini 1.00
- 2 o1 ↑1 0.85
- 3
DeepSeek-R1-Distill-Qwen-32B ↑4 0.70
- 4
Qwen3-32B ↑1 0.70
- 5
Gemini 2.5 Pro Mar 2025 ↓3 0.65
- 6 Claude 3.7 Sonnet ↓2 0.50
- 7
Qwen2.5-72B ↑4 0.20
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.20
- 9
DeepSeek-V3 Mar 2025 ↓3 0.20
- 10
Doubao-1.5-pro 0.15
Numeric Bricks
Raw game score. 12 models we track have a score here.
- 1
Qwen2.5-72B ↑10 0.00
- 2 o1 ↑1 0.00
- 3
DeepSeek-R1-Distill-Qwen-32B ↑4 0.00
- 4
DeepSeek-R1-Distill-Qwen-7B ↑8 0.00
- 5
Gemini 2.0 Flash Thinking Jan 2025 ↑3 0.00
- 6
Doubao-1.5-pro ↑4 0.00
- 7 o3-mini ↓6 0.00
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.00
- 9 Claude 3.7 Sonnet ↓5 0.00
- 10
DeepSeek-V3 Mar 2025 ↓4 0.00
Minesweeper
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.74
- 2 o3-mini ↓1 0.56
- 3 Claude 3.7 Sonnet ↑1 0.30
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.23
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.16
- 6
Qwen3-32B ↓1 0.12
- 7
DeepSeek-V3 Mar 2025 ↓1 0.06
- 8
DeepSeek-R1-Distill-Qwen-7B ↑4 0.04
- 9
Gemini 2.0 Flash Feb 2025 0.01
- 10
Gemini 2.0 Flash Thinking Jan 2025 ↓2 0.00
Nullify
Raw game score. 12 models we track have a score here.
- 1
Qwen3-32B ↑4 0.35
- 2 o1 ↑1 0.30
- 3
DeepSeek-R1-Distill-Qwen-32B ↑4 0.30
- 4
Gemini 2.0 Flash Feb 2025 ↑5 0.30
- 5
DeepSeek-V3 Mar 2025 ↑1 0.30
- 6
Gemini 2.5 Pro Mar 2025 ↓4 0.30
- 7
Qwen2.5-72B ↑4 0.25
- 8
Gemini 2.0 Flash Thinking Jan 2025 0.25
- 9
Doubao-1.5-pro ↑1 0.25
- 10 o3-mini ↓9 0.25
PVZ
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 61.3
- 2 o1 ↑1 54.9
- 3 o3-mini ↓2 41.2
- 4 Claude 3.7 Sonnet 40.3
- 5
Qwen3-32B 37.4
- 6
DeepSeek-R1-Distill-Qwen-32B ↑1 31.8
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 26.0
- 8
Qwen2.5-72B ↑3 24.9
- 9
DeepSeek-V3 Mar 2025 ↓3 20.4
- 10
Gemini 2.0 Flash Feb 2025 ↓1 19.6
Long Cat
Raw game score. 12 models we track have a score here.
- 1 o1 ↑2 0.84
- 2 o3-mini ↓1 0.70
- 3
Qwen3-32B ↑2 0.62
- 4
Gemini 2.5 Pro Mar 2025 ↓2 0.60
- 5 Claude 3.7 Sonnet ↓1 0.32
- 6
DeepSeek-V3 Mar 2025 0.30
- 7
DeepSeek-R1-Distill-Qwen-32B 0.22
- 8
Gemini 2.0 Flash Feb 2025 ↑1 0.10
- 9
Gemini 2.0 Flash Thinking Jan 2025 ↓1 0.08
- 10
Qwen2.5-72B ↑1 0.04
Black White Copy
Raw game score. 12 models we track have a score here.
- 1
Gemini 2.5 Pro Mar 2025 ↑1 0.38
- 2 o1 ↑1 0.36
- 3 o3-mini ↓2 0.36
- 4
Qwen3-32B ↑1 0.34
- 5
DeepSeek-R1-Distill-Qwen-32B ↑2 0.20
- 6 Claude 3.7 Sonnet ↓2 0.18
- 7
Gemini 2.0 Flash Thinking Jan 2025 ↑1 0.10
- 8
DeepSeek-V3 Mar 2025 ↓2 0.08
- 9
Gemini 2.0 Flash Feb 2025 0.02
- 10
Qwen2.5-72B ↑1 0.00
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official KORGym leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official KORGym 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 Games and long tasks
- ARC-AGI-3 In use Interactive game environments where agents must explore, infer goals and rules, and act efficiently without instructions. GPT-6.2 Astra 100%
- Orak Fading LLM agents play 12 popular video games across most genres through an MCP interface. Each game scores in-game progress or outcomes, and the board ranks models by their average rank across the games.
Gemini 2.5 Pro 3.90
- MineDojo Fading Agents complete open-ended tasks in Minecraft, such as mining blocks and fighting mobs, from a large simulation suite. Programmatic checks score many tasks.
Seed2.0 Pro 49%
- TextQuests Only in system cards Agents play 25 classic interactive fiction games. The score is harmful actions per 100 steps, where lower is better. The board also reports game progress.
Llama 4 Scout 7.00
- GameWorld Only in system cards Multimodal game agents play 34 browser games across 170 tasks. Outcomes are checked from the game state. GPT-6.1 Terra 76.8%
- Swarm Bench Only in system cards Models orchestrate agent swarms on complex tasks that benefit from coordinated decomposition and parallel execution. Kimi K3 76.3%
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 Qwen3-32B. 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 (18 entries).
- Pass Models scored 12 models we track, plus 6 other entries on its leaderboard.
- Warning Newest score The newest came 19 months ago, in April 2025, for Qwen3-32B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 2.5 Pro (#104).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on razor233.github.io: 13 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.