Released: 2026-06-28
google/gemini-3-flash-preview
4
3
1
2
9
7
6
5
8
2
7
6
3
4
8
1
9
5
8
9
5
6
5
1
4
3
2
3
4
9
8
5
2
6
7
1
1
5
7
9
6
4
2
8
3
6
2
8
1
7
3
5
4
9
7
6
2
5
3
1
9
8
4
5
8
3
4
4
9
7
1
2
9
1
4
7
2
5
8
6
3
openai/gpt-4o
openai
5
6
1
2
9
7
4
3
8
2
7
4
3
4
8
1
9
5
8
9
3
6
5
1
2
0
7
3
4
5
8
6
2
0
7
1
9
5
7
1
3
4
6
8
2
6
2
8
5
7
9
3
4
0
1
3
2
7
8
0
9
5
4
4
8
9
0
1
0
7
2
0
7
0
6
9
2
5
8
6
0
mistralai/mistral-medium-3
mistral
5
3
6
2
9
7
4
1
8
1
7
8
3
4
6
5
9
2
8
9
4
5
6
1
3
2
7
3
4
7
8
5
6
2
7
1
2
5
1
7
3
4
6
8
9
6
2
9
1
8
3
7
4
5
7
8
2
6
1
5
9
3
4
4
1
3
9
7
9
7
5
6
9
6
5
4
2
8
1
6
3
meta-llama/llama-3.3-70b-instruct
meta
5
8
1
2
9
7
4
6
3
7
4
3
3
4
5
2
9
1
8
6
9
1
5
1
7
3
2
3
4
5
8
2
6
1
7
9
9
5
7
4
3
8
6
2
1
6
1
2
9
7
4
8
5
3
4
9
8
5
1
3
9
1
6
2
7
6
1
8
9
5
4
8
1
3
4
6
5
2
3
8
9
deepseek/deepseek-chat-v3-0324
deepseek
1
3
4
2
9
7
8
5
6
5
7
6
3
4
8
1
9
2
8
2
9
5
6
1
4
3
7
3
4
8
6
5
2
9
7
1
2
5
7
9
1
4
6
8
3
6
9
1
7
8
3
2
4
5
7
6
2
1
3
5
9
8
4
4
8
3
8
7
9
5
2
1
9
1
5
4
2
6
3
6
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.