Released: 2026-07-29
google/gemini-3-flash-preview
5
3
4
8
2
7
6
9
1
9
6
7
3
1
2
5
8
4
1
2
8
6
5
4
3
7
9
2
8
5
1
6
3
9
4
7
4
1
9
7
3
8
2
5
6
6
7
3
5
9
4
1
2
8
8
2
6
9
7
5
4
3
1
3
9
2
4
8
6
7
1
5
7
5
1
2
4
9
8
6
3
openai/gpt-4o
openai
8
5
4
3
6
7
2
9
1
9
6
3
1
4
2
5
8
7
1
2
7
6
5
8
3
4
9
6
8
5
7
1
3
9
2
4
4
3
9
8
2
6
7
5
6
2
7
1
5
9
4
1
3
8
7
1
8
9
3
5
4
6
2
3
9
2
4
8
1
6
7
5
5
4
6
2
7
9
8
1
3
mistralai/mistral-medium-3
mistral
5
3
4
8
6
7
2
9
1
9
6
8
1
4
2
5
3
7
1
2
7
6
5
9
3
8
4
2
8
5
3
1
4
6
7
9
4
1
3
9
7
8
2
5
6
6
7
9
5
2
3
1
4
8
8
4
6
7
3
5
4
2
9
3
9
2
4
8
6
7
1
5
7
5
1
2
9
3
8
6
4
meta-llama/llama-3.3-70b-instruct
meta
5
8
4
3
2
7
1
9
6
6
1
3
5
4
9
8
7
2
7
2
9
1
8
6
3
5
4
8
5
6
4
3
1
2
7
9
4
9
1
7
6
8
5
3
2
2
7
8
9
1
5
4
6
3
3
6
5
2
7
4
9
1
8
1
4
2
8
9
3
7
6
5
9
3
7
6
5
2
8
4
1
deepseek/deepseek-chat-v3-0324
deepseek
8
5
3
4
6
7
2
9
1
9
6
4
8
3
2
5
7
0
1
2
7
6
5
9
3
8
4
6
1
9
5
2
3
7
4
8
4
3
2
7
1
8
9
5
6
5
7
8
4
9
6
1
2
3
7
8
6
3
4
5
4
6
2
3
9
5
9
8
4
6
1
7
2
4
1
2
7
1
8
3
9
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.