Released: 2026-06-10
google/gemini-3-flash-preview
3
5
7
8
1
9
4
6
2
8
1
4
6
2
7
3
9
5
2
6
9
5
3
4
1
8
7
1
4
6
3
7
2
9
5
8
5
2
3
9
8
1
7
4
6
7
9
8
4
5
6
3
2
1
6
8
1
2
9
5
7
4
3
4
7
2
1
6
3
5
8
9
9
3
5
4
7
8
2
1
6
openai/gpt-4o
openai
5
7
3
9
1
4
8
6
2
8
1
4
6
3
7
2
9
5
9
6
2
5
8
0
1
7
4
1
4
6
3
2
0
9
5
7
3
2
9
8
7
5
4
1
6
7
5
8
1
4
6
3
2
9
6
8
1
7
5
9
9
4
3
4
9
5
2
6
3
7
8
1
2
3
7
4
9
8
5
0
5
meta-llama/llama-3.3-70b-instruct
meta
5
8
9
3
1
7
2
6
4
8
3
4
9
5
2
1
7
6
2
6
1
5
4
8
3
9
7
9
5
7
3
6
4
8
2
1
4
2
3
1
9
5
6
8
7
7
1
6
8
2
3
5
4
9
6
9
5
7
8
1
4
3
2
3
4
2
6
7
9
1
5
8
1
7
8
4
3
6
9
2
5
mistralai/mistral-medium-3
mistral
3
9
5
8
1
4
7
6
2
8
2
4
6
3
7
5
9
1
4
6
7
5
2
9
1
8
3
1
4
2
3
8
6
9
5
7
5
8
9
2
7
1
3
4
6
7
3
6
9
5
4
8
2
1
6
1
8
7
9
5
2
3
4
2
5
3
1
4
3
6
7
9
9
3
7
4
6
8
5
1
2
deepseek/deepseek-chat-v3-0324
deepseek
5
9
7
8
1
4
3
6
2
8
2
4
9
6
7
5
1
3
3
6
1
5
2
3
1
8
4
4
8
2
3
7
6
9
5
1
1
5
3
2
9
8
4
7
6
7
9
6
1
4
5
8
2
9
6
4
8
7
5
2
1
9
3
2
7
5
6
3
1
2
4
9
9
3
1
4
8
9
6
2
7
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.