Released: 2026-03-10
openai/gpt-4o
openai
2
6
1
4
3
8
7
5
9
4
5
9
1
7
6
3
8
2
3
7
8
9
5
2
1
4
6
9
4
5
7
6
1
8
2
3
8
3
2
5
4
9
7
6
1
1
6
7
8
2
3
9
7
4
7
2
6
5
8
4
5
9
1
5
1
4
3
9
5
6
3
7
6
8
3
2
1
7
4
1
5
google/gemini-3-flash-preview
2
9
6
4
5
8
7
3
1
5
4
1
6
7
3
9
8
2
8
7
3
9
1
2
5
4
6
9
8
5
7
6
1
4
2
3
3
7
2
5
4
9
1
6
8
1
6
4
8
3
2
5
7
9
7
2
8
2
9
4
6
9
1
6
1
9
3
8
5
2
4
7
4
5
7
1
2
6
3
9
8
deepseek/deepseek-chat-v3-0324
deepseek
2
6
8
4
5
3
1
7
9
3
9
1
2
7
6
5
8
4
5
7
4
9
1
8
2
3
6
9
4
5
6
8
1
7
2
3
8
3
2
5
4
9
7
6
1
1
6
7
8
2
4
9
5
0
7
2
6
0
0
0
4
9
1
4
1
9
3
6
5
8
0
7
6
5
3
7
9
2
0
1
8
mistralai/mistral-medium-3
mistral
2
3
8
4
1
6
7
5
9
4
5
6
2
7
3
1
8
9
9
7
1
9
5
8
3
2
6
9
4
5
7
8
1
6
2
3
3
8
7
5
4
9
2
6
1
1
6
2
3
9
4
8
7
5
7
2
4
8
6
3
5
9
1
5
1
9
3
2
5
4
8
7
8
9
3
6
5
7
1
4
2
meta-llama/llama-3.3-70b-instruct
meta
5
3
4
6
7
8
9
1
2
6
7
2
1
9
5
3
4
8
1
9
8
3
4
2
5
6
7
8
5
9
7
6
1
4
2
3
4
2
6
8
5
3
7
9
1
7
1
3
9
2
4
8
5
6
9
6
1
5
3
7
2
8
4
2
8
7
4
1
9
6
3
5
3
4
5
2
8
6
1
7
9
anthropic/claude-3.7-sonnet
anthropic
2
9
1
4
5
8
3
7
5
4
5
6
1
7
3
9
8
2
3
7
8
9
2
4
5
1
6
9
4
5
8
6
1
7
2
3
8
3
7
5
4
9
1
6
5
1
6
2
7
3
5
8
4
9
7
2
3
6
8
4
5
9
1
6
1
4
3
9
5
2
8
7
5
8
9
2
1
7
6
3
4
google/gemini-2.0-flash-001
2
5
9
4
1
6
3
7
8
6
3
1
5
7
2
9
8
4
8
7
4
9
3
8
5
1
6
9
4
5
7
6
1
8
2
3
3
8
7
2
4
9
1
6
5
1
6
2
8
5
3
7
4
9
7
2
3
6
8
4
5
9
1
4
1
8
3
9
5
6
2
7
5
9
6
1
2
7
4
3
8
x-ai/grok-3-mini-beta
x-ai
2
6
9
4
3
5
1
7
8
3
5
1
6
7
2
9
8
4
8
7
4
9
1
3
5
6
2
9
1
5
3
8
6
7
4
2
4
3
8
2
5
7
6
1
9
1
6
2
7
9
4
8
3
5
7
2
6
8
4
3
5
9
1
6
1
3
5
2
9
4
8
7
5
8
2
1
6
4
3
7
9
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.