Released: 2026-09-07
mistralai/mistral-medium-3
mistral
8
3
6
1
5
9
7
4
2
1
2
4
4
7
8
3
5
9
7
5
9
3
2
5
8
1
6
4
5
3
8
1
6
9
2
7
6
1
2
5
6
3
4
7
8
9
8
7
2
4
1
6
3
5
2
4
1
6
3
7
5
9
8
3
7
8
9
2
4
1
6
5
5
9
6
7
8
2
3
8
4
openai/gpt-4o
openai
8
2
6
9
1
3
5
4
7
5
3
1
4
7
6
8
2
9
4
7
9
8
2
5
3
1
6
6
5
3
7
8
4
9
6
2
9
1
4
5
6
3
2
7
8
7
8
2
2
9
1
6
5
3
2
4
8
6
3
9
7
8
1
1
9
5
3
2
8
4
6
0
3
6
7
1
5
7
1
8
4
meta-llama/llama-3.3-70b-instruct
meta
8
6
5
3
9
7
1
4
2
4
8
2
5
7
1
3
6
9
3
9
9
2
8
5
4
7
1
5
1
4
9
3
6
8
2
7
2
7
6
1
4
9
5
3
8
9
3
8
7
2
4
6
1
5
1
5
3
6
1
2
7
9
4
6
4
1
8
5
3
2
8
9
7
2
9
4
6
8
5
1
3
deepseek/deepseek-chat-v3-0324
deepseek
8
3
5
9
1
6
2
4
7
6
2
1
4
7
8
5
3
9
7
4
9
3
2
5
8
1
6
4
5
6
7
8
2
9
1
3
9
1
2
5
6
3
4
7
8
3
8
7
2
4
1
6
5
9
2
4
8
6
9
3
1
7
5
1
7
3
8
5
4
9
6
2
5
6
9
1
3
7
4
8
2
google/gemini-3-flash-preview
8
7
3
9
1
6
5
4
2
6
5
2
4
7
8
1
3
9
1
9
9
3
2
5
7
6
8
3
5
6
7
8
4
9
2
1
9
1
4
5
6
3
8
7
2
8
2
7
2
9
1
6
5
4
2
4
7
6
5
9
3
1
8
7
8
5
1
2
3
4
9
6
9
3
1
4
4
7
2
8
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.