Released: 2026-06-21
openai/gpt-4o
openai
1
9
7
8
6
3
2
5
4
4
3
5
2
9
1
8
6
7
6
8
2
4
7
5
1
9
3
3
1
4
7
8
9
5
6
2
2
4
9
6
1
5
7
3
8
7
5
8
3
2
6
4
1
9
5
7
6
9
4
2
3
1
8
8
6
1
5
3
7
9
2
5
9
2
3
1
5
8
6
4
7
google/gemini-3-flash-preview
9
5
7
8
6
3
1
2
4
4
3
2
5
9
1
6
8
7
1
6
8
4
7
2
3
9
5
3
1
6
2
4
7
5
9
8
2
4
9
6
8
5
7
3
1
7
8
5
1
3
9
4
6
2
6
7
4
3
5
2
8
1
9
8
2
1
9
4
6
7
7
5
5
9
3
7
1
8
2
4
6
mistralai/mistral-medium-3
mistral
5
1
4
8
6
3
2
5
9
6
3
2
5
9
1
8
4
7
9
8
7
4
7
2
3
9
1
1
2
3
7
8
9
5
6
4
2
4
9
3
5
5
7
3
8
7
5
8
6
1
4
4
2
3
3
7
5
1
2
2
6
1
4
8
9
1
9
4
7
1
8
5
4
6
6
2
3
8
9
7
6
meta-llama/llama-3.3-70b-instruct
meta
5
1
9
8
6
4
3
2
7
3
4
2
1
5
7
8
6
9
6
8
7
3
2
5
1
4
9
1
5
3
9
8
6
4
7
2
2
6
4
7
3
1
9
5
8
9
7
8
5
4
2
6
1
3
4
2
1
6
9
8
5
3
7
8
3
5
4
1
9
2
7
6
7
9
6
2
7
3
5
8
1
deepseek/deepseek-chat-v3-0324
deepseek
5
9
4
8
6
3
1
2
7
6
3
8
2
9
1
5
4
7
1
2
7
4
7
5
6
9
3
3
6
2
1
4
7
5
8
9
2
4
9
6
8
5
7
3
1
7
1
5
3
2
9
4
6
8
4
7
6
5
3
2
8
1
9
8
6
1
9
4
7
2
5
3
9
5
3
7
1
8
2
4
6
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.