Released: 2026-07-04
google/gemini-3-flash-preview
2
6
5
3
8
7
4
1
9
1
7
3
2
6
4
9
8
5
8
4
9
5
1
9
3
7
2
3
5
6
4
7
2
7
9
1
9
8
1
6
2
5
3
4
7
2
3
4
1
9
8
5
6
3
6
9
2
5
4
1
8
3
7
4
2
7
8
5
9
1
3
6
5
1
8
9
3
6
2
7
4
openai/gpt-4o
openai
3
6
5
9
8
7
4
1
2
1
4
7
2
5
3
9
6
8
8
9
2
6
1
4
7
5
3
5
3
6
4
7
2
8
9
1
9
7
1
3
2
5
6
8
4
2
8
4
1
6
9
5
7
3
6
5
3
8
9
1
2
4
7
4
2
9
7
3
8
1
3
5
7
1
8
5
3
6
3
2
9
mistralai/mistral-medium-3
mistral
2
6
5
3
8
7
4
1
9
1
3
7
2
4
5
9
8
6
8
9
4
1
6
3
2
7
5
3
2
6
4
5
8
7
9
1
9
4
1
7
2
9
6
5
8
7
8
9
1
3
2
5
4
6
6
1
2
5
9
4
8
3
7
4
5
8
6
7
1
3
2
9
5
7
3
9
3
6
1
8
2
meta-llama/llama-3.3-70b-instruct
meta
3
6
5
1
8
7
4
2
9
1
5
9
2
6
4
9
8
7
8
7
4
5
3
9
2
1
6
5
3
2
4
1
8
7
9
6
2
9
1
7
5
6
3
4
8
4
1
8
3
9
2
6
5
7
6
8
7
5
4
3
1
9
2
9
4
3
8
7
1
5
6
2
7
2
6
9
5
8
4
3
1
deepseek/deepseek-chat-v3-0324
deepseek
2
6
5
3
8
7
4
1
9
1
3
7
2
5
4
9
6
8
8
4
9
6
1
5
3
2
7
3
2
6
4
7
8
7
9
1
5
8
1
6
2
9
3
7
4
9
7
4
1
6
3
5
8
2
6
9
3
5
4
2
8
7
7
4
1
2
7
9
8
6
3
5
7
5
8
9
3
6
2
4
1
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.