Released: 2026-07-14
google/gemini-3-flash-preview
6
3
7
8
4
1
5
10
9
9
5
2
1
3
2
6
4
8
4
8
1
5
2
6
7
3
9
8
7
3
2
1
5
4
9
6
1
4
6
9
2
8
7
5
3
5
2
9
7
6
3
8
4
1
7
3
5
6
8
9
2
1
4
2
9
8
1
5
4
3
6
7
3
1
4
3
9
7
6
2
5
openai/gpt-4o
openai
6
8
2
7
5
1
3
4
9
1
7
9
4
3
2
6
8
5
4
5
3
9
6
8
7
1
2
8
4
7
5
1
6
2
9
3
9
6
1
2
7
8
4
5
3
5
2
6
3
9
4
8
7
1
2
3
8
6
8
5
1
9
7
7
9
5
1
4
3
9
6
8
3
1
4
8
2
7
5
2
6
mistralai/mistral-medium-3
mistral
6
5
2
4
7
1
3
8
9
1
7
9
2
3
8
6
4
5
4
3
8
5
9
6
7
1
2
8
4
5
3
6
2
1
9
7
9
6
1
9
4
8
2
5
3
7
2
3
8
1
5
4
6
9
3
8
7
6
2
4
5
9
1
2
9
6
1
5
3
8
7
4
5
1
4
7
8
9
3
2
6
meta-llama/llama-3.3-70b-instruct
meta
6
8
5
4
2
1
3
7
9
3
4
9
1
7
5
6
2
8
4
1
2
8
9
6
7
5
3
8
5
6
3
1
4
2
9
7
9
7
1
5
8
3
4
6
2
2
3
4
7
6
9
1
8
5
5
9
8
6
4
7
2
3
1
1
6
3
2
5
8
9
4
7
7
2
7
9
3
1
5
1
6
deepseek/deepseek-chat-v3-0324
deepseek
6
7
3
8
4
1
5
2
9
1
5
2
9
3
6
7
8
4
4
8
9
2
5
7
6
1
3
8
3
6
1
2
5
4
9
7
7
9
1
4
6
8
2
5
3
5
2
4
7
9
3
8
6
1
2
4
5
6
8
9
3
7
1
3
9
7
5
1
2
9
4
8
9
1
8
3
7
4
1
2
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.