Released: 2026-06-16
google/gemini-3-flash-preview
7
8
2
4
3
9
6
1
5
1
3
5
2
6
7
8
9
4
4
9
6
1
8
5
7
3
2
6
7
8
5
2
1
9
4
3
5
2
1
3
9
6
4
8
7
9
4
3
8
7
4
5
6
1
8
5
4
6
4
2
3
7
9
2
1
7
9
5
3
4
8
6
3
6
9
7
1
8
2
5
4
openai/gpt-4o
openai
7
8
5
4
3
9
6
1
2
1
3
6
2
5
7
9
8
4
4
9
2
6
8
1
7
3
5
6
7
8
5
2
4
1
9
3
5
4
1
3
9
6
2
8
7
9
2
3
1
7
8
5
6
4
8
5
4
1
6
2
3
7
9
2
1
7
9
5
3
4
2
6
3
6
9
7
1
5
8
5
1
mistralai/mistral-medium-3
mistral
5
8
3
4
2
9
6
1
7
1
6
7
2
3
8
9
5
4
4
9
2
6
8
5
7
3
1
2
7
8
5
9
1
4
6
3
3
5
1
7
4
6
2
8
9
9
4
6
8
7
3
5
2
1
8
3
4
1
6
2
5
7
9
2
1
5
9
5
7
8
4
6
7
2
9
3
1
4
6
5
8
meta-llama/llama-3.3-70b-instruct
meta
5
8
1
4
3
9
6
2
7
1
6
3
2
5
8
4
9
4
4
9
2
1
8
7
3
5
6
8
7
9
5
2
1
4
3
6
3
5
6
8
9
6
1
8
2
9
1
4
7
6
5
8
2
3
2
3
8
9
1
4
5
7
9
6
4
5
3
7
2
9
1
8
7
2
9
6
4
3
5
1
8
deepseek/deepseek-chat-v3-0324
deepseek
5
8
2
4
3
9
6
1
7
1
3
7
2
6
5
8
9
4
6
9
4
1
8
7
2
3
5
4
7
8
5
9
1
0
2
3
3
2
5
6
4
8
9
7
1
9
1
6
3
7
2
5
4
8
8
5
1
6
2
4
3
7
9
2
6
3
9
5
7
1
8
0
7
4
9
8
1
3
0
5
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.