Released: 2026-09-11
openai/gpt-4o
openai
8
5
4
2
1
3
6
7
9
2
6
7
7
9
6
4
1
5
3
9
1
5
8
7
2
6
4
6
7
5
4
5
1
9
2
8
1
4
8
9
2
6
5
3
7
9
2
3
3
7
8
7
4
1
7
1
9
8
4
5
3
5
6
5
8
6
7
3
2
8
9
1
4
3
2
6
1
9
1
8
2
google/gemini-3-flash-preview
8
5
4
2
7
3
6
1
9
2
6
1
9
4
5
3
8
5
3
9
7
5
8
6
2
1
4
6
7
9
4
5
8
1
2
3
1
4
5
9
3
2
8
6
7
8
2
3
6
1
7
7
5
4
7
3
2
8
9
1
5
4
6
5
8
6
3
2
4
7
9
1
4
1
9
7
6
5
3
2
8
mistralai/mistral-medium-3
mistral
5
7
4
2
1
3
6
8
9
8
6
2
7
4
9
1
3
5
3
9
1
5
8
6
2
4
7
6
4
7
4
5
1
9
2
3
1
5
8
9
2
7
4
6
3
9
2
3
6
4
8
7
5
1
7
1
5
8
3
2
9
4
6
5
8
6
3
7
4
1
9
2
4
3
9
1
6
5
8
7
2
deepseek/deepseek-chat-v3-0324
deepseek
7
5
4
2
1
3
6
8
9
2
8
6
7
4
9
1
3
5
3
9
1
5
8
6
2
4
7
6
7
8
4
5
2
9
1
3
1
4
5
9
6
7
8
2
4
9
2
3
1
4
8
7
6
5
4
3
9
8
2
1
5
7
6
5
8
7
6
3
4
4
9
1
8
6
2
3
7
5
4
3
2
meta-llama/llama-3.3-70b-instruct
meta
5
3
4
2
8
3
6
1
9
1
7
8
5
4
9
2
3
5
3
9
2
5
8
1
2
6
7
6
4
3
4
5
8
7
9
1
1
5
9
9
7
6
8
4
2
8
2
3
1
9
5
4
7
6
9
1
7
8
3
4
5
2
8
5
8
6
7
2
3
9
1
4
4
9
1
6
8
7
3
5
2
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.