Released: 2026-07-26
google/gemini-3-flash-preview
6
5
4
2
8
7
1
3
9
3
7
2
6
9
5
8
1
4
9
1
8
4
1
3
2
5
7
5
9
7
8
2
4
3
6
1
8
6
3
5
1
9
7
4
2
4
2
1
7
3
6
5
9
8
5
3
9
9
4
8
6
7
1
1
4
5
3
7
6
9
8
2
7
8
6
1
2
9
4
3
5
openai/gpt-4o
openai
5
6
4
8
2
7
1
3
9
3
7
1
6
9
5
8
4
2
2
9
8
4
1
3
5
5
7
4
8
3
7
5
2
9
6
1
8
5
6
3
4
9
7
2
7
7
2
9
5
3
1
4
8
6
9
3
7
2
8
4
6
1
8
1
4
5
9
7
6
2
9
3
6
1
2
1
6
8
3
7
5
meta-llama/llama-3.3-70b-instruct
meta
5
8
4
6
7
7
3
2
9
3
1
6
8
9
5
8
4
2
9
5
8
4
1
2
6
5
7
7
4
3
2
5
8
1
6
1
8
9
1
3
4
9
7
8
6
2
6
5
5
3
1
4
9
8
1
3
9
9
2
6
5
7
4
4
2
5
1
8
3
9
1
3
6
7
2
9
6
4
8
3
5
mistralai/mistral-medium-3
mistral
5
6
4
8
2
7
3
1
9
7
1
3
2
9
4
8
6
5
9
2
8
4
6
3
1
5
7
4
5
7
3
8
2
9
6
1
8
3
6
1
5
9
7
4
2
6
2
9
5
3
7
4
8
1
2
3
1
9
7
8
6
5
4
1
4
5
6
2
6
5
3
8
3
8
6
1
4
5
2
7
9
deepseek/deepseek-chat-v3-0324
deepseek
2
5
4
8
6
7
1
3
9
3
1
6
2
9
5
8
4
7
7
9
8
4
1
3
2
5
6
5
7
9
6
4
2
3
6
1
8
4
3
6
2
9
7
1
5
6
2
1
5
3
8
4
7
9
9
3
2
7
5
4
6
8
1
1
8
5
9
7
6
4
2
3
4
6
7
1
8
2
9
3
5
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.