Released: 2026-04-02
openai/gpt-4o
openai
3
8
7
9
4
6
2
1
5
9
6
2
1
5
3
4
7
8
1
4
5
6
7
2
9
3
0
8
5
9
7
3
1
0
4
6
7
2
6
4
8
9
3
5
1
4
1
3
5
2
8
6
9
7
5
7
1
3
9
7
8
2
4
6
9
4
8
1
5
7
0
3
2
3
8
5
6
4
7
0
9
meta-llama/llama-3.3-70b-instruct
meta
3
8
7
2
4
5
1
9
6
6
5
2
9
8
3
4
7
1
1
4
9
6
7
2
8
5
3
8
5
3
1
9
6
7
2
4
4
2
1
4
5
8
3
6
9
9
7
6
5
3
1
2
8
4
5
1
4
7
2
9
6
3
8
2
9
8
4
1
7
5
6
3
7
3
5
8
6
4
9
1
2
anthropic/claude-3.7-sonnet
anthropic
3
1
7
9
4
8
5
6
2
5
6
2
1
7
3
9
4
8
9
4
8
6
5
2
1
7
3
8
5
9
7
2
1
3
4
6
6
2
4
4
8
5
9
3
1
7
8
3
2
1
6
4
9
5
4
3
1
6
9
7
8
2
5
1
9
6
8
3
4
2
5
7
2
7
5
5
6
9
7
1
4
google/gemini-2.0-flash-001
3
1
7
9
4
5
6
8
2
9
6
2
1
8
3
4
7
5
5
4
8
6
7
2
1
3
9
8
5
4
7
3
1
2
9
6
7
9
6
4
2
8
3
5
1
1
2
3
5
6
9
8
4
7
4
8
1
3
9
7
5
2
6
6
7
5
8
1
4
9
3
2
2
3
9
5
6
1
7
9
4
google/gemini-3-flash-preview
3
8
7
1
4
5
2
6
9
5
6
2
9
8
3
1
7
4
1
4
9
6
7
2
5
3
8
8
5
4
7
2
1
3
1
6
9
2
6
4
3
8
5
7
1
7
1
3
5
6
9
4
9
2
4
5
1
3
9
7
8
2
6
6
7
5
8
1
2
9
4
3
2
9
8
5
4
6
7
1
5
mistralai/mistral-medium-3
mistral
3
9
7
1
4
8
5
6
2
5
6
2
7
9
3
1
4
8
1
4
8
6
5
2
9
3
7
8
5
9
2
3
1
4
7
6
6
2
4
4
7
9
3
8
1
7
1
3
8
6
4
2
9
5
9
8
1
3
9
7
6
2
4
4
3
6
8
2
6
7
1
9
2
7
5
5
8
9
7
4
3
deepseek/deepseek-chat-v3-0324
deepseek
3
1
7
9
4
5
2
6
8
5
6
2
7
8
3
1
4
9
9
4
8
6
1
2
5
3
7
8
5
9
2
3
1
4
7
6
7
2
6
4
5
9
8
3
1
1
7
3
8
6
4
2
9
5
6
3
1
4
9
7
8
2
5
4
9
5
8
2
6
3
1
7
2
8
4
5
3
1
7
6
9
x-ai/grok-3-mini-beta
x-ai
3
1
7
9
4
6
5
8
2
9
6
2
1
8
3
4
7
5
5
4
8
6
7
2
1
3
9
8
5
9
7
3
1
2
4
6
7
2
6
4
5
8
3
9
1
1
3
4
2
6
9
8
5
7
6
8
1
3
9
7
5
2
4
4
9
5
8
2
6
7
1
3
2
7
3
5
1
4
9
6
8
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.