Released: 2026-08-07
openai/gpt-4o
openai
2
8
9
1
4
6
7
5
3
4
7
5
9
3
8
2
1
6
1
6
3
8
2
5
9
4
7
5
3
9
6
1
4
7
2
8
6
4
7
2
5
8
1
3
9
8
9
2
7
9
3
6
1
4
9
1
4
5
3
2
8
7
6
3
2
6
5
7
1
4
4
9
7
5
8
4
6
9
3
8
1
google/gemini-3-flash-preview
9
7
2
3
4
6
8
1
5
5
8
4
9
1
7
2
6
3
1
6
3
8
2
5
9
4
7
5
4
9
1
6
2
7
3
8
2
3
6
4
8
7
1
5
9
8
7
1
5
9
3
6
2
4
4
1
8
6
3
9
5
7
2
3
2
9
5
7
1
8
4
6
7
5
6
2
4
8
3
9
1
mistralai/mistral-medium-3
mistral
3
7
4
6
4
5
2
8
9
8
9
2
3
7
1
4
6
5
1
6
5
8
2
9
9
3
7
5
3
9
1
6
4
7
2
8
2
4
6
7
5
8
1
9
3
7
8
1
2
9
3
6
5
4
6
1
7
9
3
2
8
7
2
4
2
8
5
1
6
3
4
6
7
5
3
4
8
7
5
1
1
deepseek/deepseek-chat-v3-0324
deepseek
2
8
3
9
4
5
7
1
6
4
7
9
1
6
8
5
3
2
1
6
5
8
2
7
9
4
3
5
3
9
6
7
4
2
1
8
6
4
2
7
5
1
8
9
3
8
1
7
2
9
3
6
5
4
9
1
6
4
3
2
8
7
5
3
2
8
5
1
6
7
4
9
7
5
4
3
8
9
2
6
1
meta-llama/llama-3.3-70b-instruct
meta
5
3
4
6
4
8
2
1
9
6
8
7
3
5
2
4
9
1
1
6
9
8
2
5
3
7
4
5
4
9
1
7
6
8
3
2
2
9
8
4
1
9
1
5
6
3
1
5
9
9
3
6
8
4
9
1
3
5
3
1
7
2
8
4
2
6
8
6
4
5
3
7
7
5
1
2
8
7
9
6
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.