Released: 2026-06-03
google/gemini-3-flash-preview
3
8
2
7
4
5
6
1
9
1
4
7
6
9
2
3
5
8
5
6
9
2
8
3
4
7
1
6
7
5
3
2
1
4
8
9
9
1
4
0
5
8
7
3
2
2
8
3
4
7
9
1
6
5
7
9
1
8
6
4
5
2
3
4
5
6
1
3
2
7
9
8
8
3
8
5
1
7
2
4
6
openai/gpt-4o
openai
3
2
6
7
4
5
8
1
9
8
4
1
6
9
2
3
5
7
5
7
9
8
1
3
2
4
6
6
8
5
3
7
9
4
2
1
9
1
7
5
2
8
6
3
4
2
3
4
1
6
4
5
6
8
7
9
2
4
8
6
1
8
3
4
6
3
9
5
1
7
9
2
1
5
8
2
3
7
9
7
5
mistralai/mistral-medium-3
mistral
8
3
4
7
6
5
2
1
9
2
1
7
6
9
4
3
8
5
5
6
9
2
8
3
1
4
7
3
2
6
1
7
9
4
5
8
9
4
1
3
5
8
6
7
2
6
8
3
4
2
7
9
5
1
7
9
2
8
4
6
5
3
3
4
5
6
9
3
1
7
2
8
1
7
8
5
2
3
6
4
9
deepseek/deepseek-chat-v3-0324
deepseek
2
8
4
7
6
5
3
1
9
3
7
9
6
1
4
2
8
5
5
1
6
2
8
3
9
4
7
1
2
7
9
3
6
4
5
8
9
6
4
1
5
8
7
3
2
8
5
3
4
7
2
1
6
9
7
9
1
8
2
4
5
6
3
4
3
2
5
6
1
7
9
8
6
4
8
3
9
7
6
2
1
meta-llama/llama-3.3-70b-instruct
meta
3
8
6
1
4
5
2
1
9
2
5
9
6
7
4
8
3
1
5
1
4
2
8
3
9
6
7
1
9
7
5
2
6
4
8
3
9
6
3
8
5
8
1
4
2
8
4
3
4
1
2
6
5
9
7
9
1
3
6
9
5
2
8
4
2
8
9
3
1
7
9
6
6
3
5
7
9
7
1
8
4
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.