Released: 2026-08-02
google/gemini-3-flash-preview
4
1
2
8
7
6
9
5
3
8
7
9
1
3
4
2
6
5
3
5
6
2
9
5
8
7
4
2
6
7
4
8
3
1
9
5
1
9
4
5
2
8
7
3
6
5
8
3
6
1
7
4
2
9
7
2
5
3
4
1
6
9
8
9
4
1
7
6
2
5
8
3
6
3
8
9
5
5
4
1
2
openai/gpt-4o
openai
8
1
2
5
7
6
9
4
3
4
7
5
1
9
4
2
8
6
3
9
6
8
5
2
1
7
4
6
2
7
4
3
9
8
1
5
1
9
4
7
2
8
5
3
6
5
8
3
6
1
7
4
2
9
7
6
1
9
4
3
5
2
8
2
5
9
3
8
1
6
4
7
9
4
8
2
6
5
7
9
1
mistralai/mistral-medium-3
mistral
4
1
2
9
7
6
8
5
3
5
3
7
1
8
4
2
9
6
3
8
6
2
5
5
9
7
4
6
2
9
4
3
7
1
8
5
1
9
4
5
2
8
6
3
7
7
8
3
6
1
9
4
2
5
7
4
5
3
6
1
2
9
8
2
6
1
8
9
2
7
4
3
9
5
8
7
4
3
4
6
1
meta-llama/llama-3.3-70b-instruct
meta
5
1
2
8
7
3
9
6
4
6
3
9
1
5
4
8
2
7
3
8
6
9
2
1
5
7
4
9
5
1
4
8
6
3
7
2
1
9
4
3
2
8
6
5
7
7
2
8
5
1
9
4
3
6
2
6
5
7
4
3
1
9
8
8
4
3
6
9
2
7
1
5
4
7
9
2
6
5
8
1
3
deepseek/deepseek-chat-v3-0324
deepseek
4
1
2
5
7
6
8
9
3
8
5
7
1
9
4
2
6
1
3
9
6
2
8
5
1
7
4
2
6
7
4
3
9
6
8
5
1
9
4
6
2
8
7
3
2
5
8
3
7
1
2
6
4
9
7
2
1
3
4
6
9
5
8
9
3
5
8
6
2
3
1
7
6
4
8
9
5
7
4
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.