Released: 2026-06-29
openai/gpt-4o
openai
3
4
6
9
8
5
2
7
1
9
8
1
2
4
6
3
5
0
7
0
5
1
3
0
8
4
6
2
7
8
4
0
1
9
3
5
5
3
4
0
6
9
7
2
8
0
1
9
7
5
2
0
6
0
1
2
0
6
9
7
4
8
3
6
5
3
8
2
1
0
9
7
8
9
7
3
0
4
6
1
2
mistralai/mistral-medium-3
mistral
2
4
6
9
8
5
1
7
3
9
8
1
2
3
4
6
5
7
7
3
5
1
6
9
8
4
2
5
6
8
4
1
3
9
2
7
4
3
2
5
9
8
7
1
6
1
7
9
6
2
5
4
8
3
1
2
3
6
4
7
8
9
5
6
5
8
9
7
1
2
3
4
9
7
4
3
5
2
6
1
8
google/gemini-3-flash-preview
2
6
4
9
8
3
1
7
5
9
8
1
2
7
4
3
5
6
5
7
3
1
6
5
8
4
9
7
1
8
4
5
2
9
3
6
4
3
5
8
9
6
7
2
1
2
6
9
7
3
1
5
8
4
1
2
3
6
4
7
8
9
5
6
5
4
9
2
1
7
3
8
8
9
7
3
5
4
6
1
2
deepseek/deepseek-chat-v3-0324
deepseek
4
6
2
9
8
3
1
7
5
5
8
1
2
7
4
3
6
9
9
7
3
1
5
6
2
4
8
2
1
8
4
6
5
9
3
7
7
3
5
8
1
9
4
2
6
6
4
9
7
3
2
5
8
1
1
2
4
6
9
7
8
5
3
6
5
3
8
2
1
7
9
4
8
9
7
3
4
5
6
1
2
meta-llama/llama-3.3-70b-instruct
meta
5
1
9
9
8
6
4
7
2
8
8
1
5
3
2
6
1
9
4
6
5
1
9
8
3
4
1
9
5
8
4
1
3
9
3
6
3
3
6
2
5
4
8
2
5
2
4
9
7
6
5
1
8
3
1
2
3
6
4
7
5
9
8
6
5
4
8
2
1
7
6
3
7
9
7
3
6
9
2
5
4
Each model receives the same Sudoku puzzle with a partially filled grid. Models are tasked with solving the puzzle by filling in all empty cells following Sudoku rules, and their solutions are evaluated without feedback or retries.