Released: 2026-09-14
openai/gpt-4o
openai
1
6
2
9
7
3
5
4
8
9
8
4
1
2
6
3
5
7
7
5
6
8
4
2
1
9
3
3
7
8
6
5
1
4
6
2
4
9
5
3
1
8
6
7
1
6
1
9
8
3
4
7
8
5
5
4
1
7
6
9
8
2
4
8
2
7
5
3
1
9
3
6
3
3
8
4
9
7
2
1
6
google/gemini-3-flash-preview
7
1
3
9
5
2
8
6
4
9
8
4
1
7
6
2
5
3
2
5
6
3
4
8
1
9
7
3
7
8
6
9
5
4
1
2
4
6
5
2
1
3
9
7
8
1
9
2
8
7
4
3
4
5
5
4
9
7
3
1
6
2
8
6
2
7
5
8
9
4
3
1
8
3
1
4
2
7
5
5
6
mistralai/mistral-medium-3
mistral
4
1
3
9
5
2
6
8
7
7
8
4
3
6
9
5
1
2
9
5
6
7
8
4
1
2
3
3
7
1
6
4
8
9
5
2
2
9
8
5
1
3
4
7
6
6
4
5
2
9
7
3
8
1
5
3
2
1
7
6
8
4
9
1
2
7
5
3
8
6
9
4
8
6
9
4
2
5
7
3
1
meta-llama/llama-3.3-70b-instruct
meta
5
3
4
9
8
2
6
7
1
8
7
2
5
3
6
1
4
9
9
1
6
7
4
5
8
3
2
3
5
9
1
2
8
4
6
7
1
4
8
6
5
3
2
9
7
7
2
5
4
9
1
3
8
6
6
9
1
8
7
4
5
2
3
4
8
3
2
6
9
7
1
5
2
6
7
3
1
5
9
4
8
deepseek/deepseek-chat-v3-0324
deepseek
1
3
2
9
5
8
6
4
7
7
8
4
1
2
6
3
5
9
9
5
6
3
7
4
1
9
8
3
7
9
6
8
5
4
1
2
4
6
5
2
1
9
8
7
3
2
1
8
8
3
4
5
6
9
5
9
1
7
6
3
2
8
4
6
2
7
5
4
1
9
3
8
8
4
3
4
9
2
7
5
6
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.