Released: 2026-08-20
openai/gpt-4o
openai
1
4
2
5
3
6
7
9
8
5
7
6
4
1
9
2
8
3
9
3
8
2
7
8
4
1
6
7
5
1
6
2
3
8
4
9
2
8
3
7
5
4
6
9
1
4
6
9
8
9
5
3
2
7
3
1
4
9
8
9
5
7
2
8
2
5
1
6
2
9
3
4
6
9
7
3
4
7
1
5
8
google/gemini-3-flash-preview
1
2
4
9
5
8
7
6
3
5
7
3
4
1
6
2
8
9
9
8
6
2
3
7
4
1
5
8
9
1
3
7
4
5
2
6
4
3
7
8
2
5
6
9
1
2
6
5
1
9
5
3
4
8
3
5
4
6
8
9
1
7
2
7
4
5
1
6
2
9
3
8
6
1
2
7
4
3
8
5
9
mistralai/mistral-medium-3
mistral
1
2
3
6
7
8
5
4
9
5
7
6
4
1
9
2
8
3
9
4
8
2
5
3
4
6
7
7
8
1
3
6
4
9
2
5
2
3
4
5
8
7
6
9
1
4
6
9
1
2
5
3
7
8
3
5
7
8
9
1
2
4
6
6
9
5
7
3
2
8
1
4
8
1
2
9
4
6
7
5
3
meta-llama/llama-3.3-70b-instruct
meta
1
5
3
7
9
6
8
4
2
7
4
2
1
5
8
6
3
9
9
8
6
2
3
4
5
7
1
5
1
9
8
2
7
3
6
4
2
3
4
9
1
5
7
8
6
8
6
1
4
7
3
2
5
9
3
9
7
5
8
2
1
4
6
4
2
5
6
4
9
8
1
3
6
7
8
3
4
1
9
2
5
deepseek/deepseek-chat-v3-0324
deepseek
1
2
3
5
6
4
7
8
9
5
7
6
4
1
9
2
8
3
9
4
8
2
3
7
4
6
5
4
3
1
6
2
8
5
7
9
2
5
7
3
4
6
8
9
1
8
6
9
7
5
1
3
2
4
3
1
4
8
9
5
6
7
2
7
8
5
9
6
2
1
3
4
6
9
2
1
4
3
9
5
8
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.