Released: 2026-03-18
openai/gpt-4o
openai
8
2
4
7
5
3
9
1
6
7
9
5
6
1
8
3
2
4
3
6
1
4
2
9
8
5
7
2
7
9
5
3
1
4
6
8
6
8
3
2
9
7
1
4
5
4
5
1
8
6
2
7
3
9
9
1
6
3
8
4
5
7
2
1
3
8
9
7
5
6
4
2
5
4
7
9
4
6
2
8
1
google/gemini-2.0-flash-001
8
9
2
4
5
3
7
1
6
7
5
4
6
1
8
9
2
3
3
6
1
7
9
2
5
4
8
6
7
9
5
3
1
4
8
2
2
8
3
4
6
7
1
9
5
4
1
5
2
8
9
6
3
7
9
4
6
1
2
3
8
7
5
1
2
8
3
7
5
6
9
4
5
3
7
9
4
6
2
8
1
google/gemini-3-flash-preview
8
1
2
7
5
3
9
4
6
7
3
5
6
9
8
1
2
4
9
6
4
1
2
4
3
5
8
2
7
9
5
3
1
4
6
8
6
8
3
4
2
7
1
9
5
4
5
1
8
6
9
2
3
7
3
9
6
2
1
4
8
7
5
1
4
8
3
7
5
6
9
2
5
2
7
9
4
6
3
8
1
anthropic/claude-3.7-sonnet
anthropic
8
4
2
7
5
3
9
1
6
7
3
5
6
1
8
2
4
9
9
6
1
4
2
9
3
5
7
2
7
9
5
3
1
4
6
8
6
8
3
2
9
7
1
9
5
4
5
1
8
6
2
7
3
9
3
2
8
1
9
6
5
7
4
1
9
4
3
7
5
8
2
3
5
3
7
9
4
8
6
9
1
meta-llama/llama-3.3-70b-instruct
meta
8
5
2
1
5
3
4
9
6
3
9
6
6
1
8
5
2
8
1
6
4
3
9
2
7
8
5
9
7
9
5
3
1
4
6
2
2
1
3
7
6
7
1
5
5
4
3
1
2
8
6
9
3
7
6
8
5
4
2
9
3
1
4
5
4
7
8
7
5
2
6
3
5
2
7
9
4
1
8
1
1
mistralai/mistral-medium-3
mistral
8
1
2
4
5
3
7
9
6
7
3
4
6
9
8
5
2
1
9
6
5
2
7
1
8
4
3
6
7
9
1
3
2
4
8
5
2
4
3
8
6
7
1
5
9
4
5
1
3
8
9
2
3
7
3
9
7
5
2
4
6
1
8
1
8
6
7
3
5
9
2
4
5
2
8
9
4
6
3
7
1
deepseek/deepseek-chat-v3-0324
deepseek
8
2
4
1
5
3
7
9
6
7
3
5
6
9
8
1
2
4
9
6
1
7
2
4
5
8
3
2
7
9
5
1
6
4
8
5
6
8
3
4
2
7
1
6
5
4
5
1
8
6
9
2
3
7
1
9
6
3
8
2
9
4
8
3
4
8
2
7
5
6
1
9
5
2
7
9
4
6
8
5
1
x-ai/grok-3-mini-beta
x-ai
8
1
2
7
5
3
4
9
6
3
9
4
6
1
8
5
2
7
7
5
6
2
3
9
1
4
8
6
7
9
1
8
2
3
5
4
2
4
3
5
6
7
9
1
8
4
2
1
8
7
4
6
3
9
9
3
8
4
2
6
7
8
1
1
6
7
3
9
5
2
4
3
5
8
4
9
3
1
8
7
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.