Released: 2026-08-22
openai/gpt-4o
openai
9
6
3
1
8
5
7
2
4
7
1
4
5
6
2
9
3
8
5
2
8
9
3
7
4
6
1
6
8
7
3
2
1
5
4
9
4
5
1
8
9
6
3
2
7
3
9
2
7
4
8
6
1
5
8
3
6
4
7
9
2
5
3
2
7
9
6
5
4
1
8
6
1
4
5
2
3
8
9
7
5
google/gemini-3-flash-preview
9
5
6
3
8
1
7
2
4
7
1
4
5
6
2
9
3
8
3
2
8
9
4
7
5
6
1
5
8
7
6
2
4
1
9
3
4
9
1
8
5
7
3
2
6
2
6
1
7
9
3
8
4
5
8
4
3
1
7
9
2
5
6
1
7
9
4
3
6
6
8
2
6
3
5
2
1
8
4
9
7
meta-llama/llama-3.3-70b-instruct
meta
9
5
7
3
8
6
1
4
2
1
3
4
5
2
9
8
6
7
6
2
8
1
4
5
3
7
9
8
6
5
9
1
4
2
3
7
7
9
1
6
5
2
4
8
3
4
8
3
7
6
1
5
9
2
2
1
9
8
7
3
6
5
4
5
7
6
4
9
8
7
1
3
3
4
2
5
1
7
9
6
8
mistralai/mistral-medium-3
mistral
9
2
6
3
8
1
7
5
4
7
1
4
5
6
2
9
8
3
3
5
8
9
4
7
2
6
1
4
8
2
6
3
5
1
7
9
5
7
1
8
9
4
3
2
6
6
9
3
7
2
3
4
1
8
1
4
3
6
7
9
2
8
5
8
6
9
1
5
2
4
3
7
2
3
5
4
1
8
6
9
2
deepseek/deepseek-chat-v3-0324
deepseek
9
2
7
1
8
6
5
3
4
6
1
4
5
3
9
7
8
2
3
5
8
4
2
7
9
6
1
4
8
2
9
6
1
8
5
7
5
7
1
8
4
2
3
2
6
1
9
6
7
5
3
4
1
8
8
6
3
6
7
9
2
4
5
2
4
9
3
1
5
6
7
8
7
3
5
2
6
8
1
9
3
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.