Released: 2026-06-22
google/gemini-3-flash-preview
5
2
6
4
9
7
1
8
3
7
9
4
6
8
1
4
5
2
1
3
4
2
5
8
6
7
9
9
7
1
3
4
2
8
6
5
2
4
8
9
1
5
3
6
7
6
5
3
8
7
4
9
2
1
4
1
5
7
3
6
2
9
8
3
8
9
5
2
4
7
1
6
8
6
7
1
6
9
5
3
4
openai/gpt-4o
openai
5
2
6
4
9
8
7
3
1
8
9
7
6
3
1
4
5
2
1
3
4
2
7
5
6
9
8
9
7
5
1
4
2
8
6
3
2
6
8
3
1
9
5
4
7
3
4
1
8
6
7
2
5
9
4
1
2
7
5
6
9
8
3
6
5
9
5
2
3
1
7
4
7
8
3
9
8
4
5
2
6
mistralai/mistral-medium-3
mistral
5
6
7
4
9
8
1
3
2
8
9
4
3
6
1
7
5
2
2
3
1
7
5
9
6
4
8
9
7
6
3
4
2
8
5
1
2
4
8
6
1
5
9
3
7
3
5
1
9
8
7
2
6
4
4
1
2
8
3
6
9
7
5
6
8
9
5
2
4
3
1
7
7
3
5
1
9
8
5
2
6
meta-llama/llama-3.3-70b-instruct
meta
5
8
6
4
9
7
2
3
1
7
1
3
5
6
1
4
8
2
8
3
2
1
5
9
6
7
4
9
7
1
8
4
2
8
5
6
2
5
8
7
1
3
9
4
7
4
6
5
2
8
5
7
1
3
4
1
9
3
7
6
5
2
8
1
9
7
5
2
8
3
6
4
3
4
5
9
6
1
1
7
2
deepseek/deepseek-chat-v3-0324
deepseek
5
6
7
4
9
8
1
2
3
8
9
4
3
6
1
4
5
2
1
3
2
7
5
4
6
8
9
9
7
5
6
4
2
8
3
1
2
4
8
9
1
3
5
6
7
6
1
3
8
7
5
2
9
4
4
1
2
3
8
6
7
9
5
7
8
9
5
2
4
3
1
6
3
5
6
1
9
7
5
4
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.