Released: 2026-05-22
google/gemini-3-flash-preview
6
7
4
3
2
8
5
9
1
1
5
2
9
6
1
7
4
3
3
8
9
4
7
5
1
2
6
4
9
6
2
3
7
8
1
5
2
3
8
1
4
5
9
6
7
5
7
1
6
8
9
3
2
4
8
6
3
7
5
4
2
1
9
7
4
5
8
1
3
6
9
2
9
1
7
5
2
6
4
3
8
google/gemini-2.0-flash-001
5
7
4
3
2
1
6
9
8
1
2
9
6
8
5
7
4
3
3
8
6
4
7
9
5
2
1
4
9
8
6
3
7
1
2
5
2
5
3
1
4
8
9
6
7
6
1
7
5
9
2
3
8
4
8
6
2
7
5
4
3
1
9
9
3
5
8
1
2
7
6
4
7
4
1
9
6
3
2
5
8
openai/gpt-4o
openai
5
7
4
3
2
8
1
9
6
1
6
2
9
6
5
7
8
3
3
8
9
4
1
7
5
2
4
4
9
6
2
3
1
8
7
5
2
5
8
1
4
9
3
6
7
7
3
1
6
5
8
2
4
9
9
2
3
7
8
4
6
1
2
8
4
5
8
9
2
9
5
1
6
1
7
5
7
6
4
3
8
meta-llama/llama-3.3-70b-instruct
meta
5
7
4
1
2
8
3
9
6
1
3
9
9
6
5
7
8
4
3
8
6
4
5
7
2
1
9
4
9
1
3
8
6
5
7
2
2
5
3
7
4
9
1
6
8
8
6
7
2
1
4
9
5
3
9
1
5
8
7
3
4
2
6
6
4
2
5
9
1
8
3
7
7
2
8
6
3
2
6
4
1
mistralai/mistral-medium-3
mistral
6
7
4
3
2
8
5
9
1
1
5
8
9
6
2
7
4
3
3
8
9
4
5
7
1
2
6
7
9
6
2
3
1
4
8
5
2
4
3
1
4
5
9
6
8
5
1
1
8
9
4
3
7
2
9
6
8
7
5
4
2
1
3
4
2
5
8
1
3
6
9
7
8
3
7
6
2
9
5
4
1
deepseek/deepseek-chat-v3-0324
deepseek
5
7
4
3
2
8
1
9
6
1
2
3
9
6
5
7
4
8
3
8
6
4
7
1
5
2
9
4
9
2
6
8
7
3
1
5
2
5
8
1
4
3
9
6
7
6
3
1
2
5
9
8
7
4
9
6
8
7
3
4
2
1
5
7
4
5
8
1
2
6
9
3
8
1
7
5
9
6
4
3
2
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.