Released: 2026-08-03
openai/gpt-4o
openai
9
5
8
4
3
7
2
6
1
7
2
3
1
8
9
4
5
6
6
4
1
5
2
3
9
8
7
2
9
6
8
7
3
1
5
4
4
1
5
2
6
9
7
3
8
3
8
7
6
5
1
2
4
9
3
6
4
9
1
5
8
2
7
8
7
2
3
4
6
5
1
9
5
9
4
3
7
8
6
7
2
google/gemini-3-flash-preview
9
8
6
4
3
2
5
1
7
7
2
3
5
1
9
4
8
6
4
5
1
6
8
7
9
2
3
2
9
7
8
1
3
6
5
4
1
3
5
2
6
4
7
9
8
6
4
8
7
5
1
2
3
9
3
2
9
1
7
5
8
4
6
8
7
2
9
4
6
3
5
1
5
1
4
3
9
8
6
7
2
mistralai/mistral-medium-3
mistral
6
8
5
4
1
2
7
3
9
7
2
3
8
3
9
4
6
1
4
9
1
6
7
5
9
2
8
2
6
9
8
9
3
6
5
4
1
5
8
2
6
4
3
9
7
9
3
4
5
5
7
2
8
6
3
4
2
9
8
1
5
6
7
8
7
6
7
2
6
8
4
3
5
1
7
3
4
8
1
7
2
meta-llama/llama-3.3-70b-instruct
meta
5
8
9
4
1
6
3
2
7
7
2
3
9
5
9
4
8
1
1
4
6
7
3
2
9
5
8
2
5
4
8
7
3
1
9
6
9
1
7
2
6
5
8
4
3
8
3
5
1
9
4
6
7
2
3
9
1
5
8
7
2
6
4
4
7
2
6
1
8
5
3
9
6
9
8
3
4
1
7
2
5
deepseek/deepseek-chat-v3-0324
deepseek
6
9
8
4
3
5
2
1
7
7
2
3
6
8
9
4
5
1
4
5
1
7
2
6
9
8
3
2
6
9
8
7
3
1
5
4
1
3
7
2
6
4
8
9
5
8
4
5
1
5
7
6
3
9
3
1
2
9
4
8
5
7
6
8
7
4
5
9
6
3
2
1
5
6
9
3
1
8
4
7
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.