Released: 2026-07-01
google/gemini-3-flash-preview
8
4
3
2
6
7
9
1
5
1
6
5
9
8
3
7
4
2
9
7
2
1
4
5
3
6
8
3
2
6
8
5
4
1
9
7
7
5
4
6
1
9
8
3
6
8
9
1
3
7
2
4
5
6
6
3
9
5
2
8
2
7
4
5
1
7
9
3
4
6
8
2
4
8
2
7
1
6
5
9
3
openai/gpt-4o
openai
8
5
4
6
7
3
9
1
2
1
6
7
5
8
2
3
4
9
9
3
2
1
4
9
7
5
8
2
4
6
8
5
7
1
3
9
7
8
3
2
9
1
5
6
4
5
9
1
3
6
4
8
2
7
3
1
5
4
2
8
6
7
9
6
2
8
9
3
5
4
9
1
4
7
9
7
1
6
2
8
3
meta-llama/llama-3.3-70b-instruct
meta
8
5
4
7
3
2
9
1
6
6
2
9
1
8
5
7
4
3
3
1
2
6
4
9
8
5
7
5
8
7
8
5
1
1
6
4
7
4
3
2
9
6
5
8
6
2
9
1
3
7
4
4
3
5
9
6
8
5
1
7
2
7
4
1
7
5
9
6
8
3
2
9
4
3
6
4
2
1
6
9
5
mistralai/mistral-medium-3
mistral
8
4
5
6
7
3
9
1
2
1
6
3
2
8
9
7
4
5
9
7
2
4
1
5
3
6
8
2
3
6
8
5
4
1
9
7
7
1
4
9
2
3
8
5
6
5
9
1
3
6
7
4
2
8
4
5
9
1
3
8
2
7
6
6
2
8
9
3
7
5
4
1
3
8
7
5
4
6
5
8
3
deepseek/deepseek-chat-v3-0324
deepseek
8
5
4
6
2
7
9
1
3
1
6
7
5
8
3
6
4
2
9
3
2
1
4
9
8
5
7
4
2
6
8
5
9
1
3
7
7
8
3
2
1
4
5
9
6
5
9
1
3
7
6
4
2
8
3
4
8
4
6
5
2
7
9
2
7
5
9
3
8
7
6
4
6
1
9
7
2
6
5
8
3
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.