Released: 2026-02-08
openai/gpt-4o
openai
4
6
5
9
2
8
3
1
7
3
1
7
5
4
2
5
9
6
8
9
2
7
3
6
4
5
1
5
8
6
2
7
1
9
4
3
1
7
3
6
9
4
8
2
5
9
2
4
3
5
8
1
7
5
7
4
9
8
6
3
5
1
2
6
3
8
1
5
7
9
8
1
2
7
1
4
8
9
6
3
4
google/gemini-2.0-flash-001
4
6
5
9
1
8
3
7
2
3
9
7
2
4
5
5
1
6
8
1
2
7
3
6
4
5
9
5
8
6
5
9
1
2
4
7
7
3
1
6
5
4
8
9
2
9
2
4
3
8
7
1
6
5
1
2
9
4
6
3
7
8
4
6
4
3
8
7
2
9
5
1
2
7
8
1
5
9
6
3
4
mistralai/mistral-medium-3
mistral
4
1
5
9
2
8
3
6
7
3
9
7
1
4
3
5
8
6
8
6
2
7
3
6
4
1
9
7
8
3
6
5
1
2
4
8
2
5
6
8
9
4
7
3
1
9
1
4
3
7
2
1
6
5
5
2
1
4
6
9
8
7
3
6
3
8
5
1
7
9
2
4
1
7
9
2
8
3
6
5
1
anthropic/claude-3.7-sonnet
anthropic
4
1
2
9
5
8
3
6
7
3
9
7
1
4
2
5
8
6
8
5
6
7
3
6
2
1
9
7
8
3
5
9
1
4
2
6
2
6
9
8
7
4
5
3
1
9
4
5
3
6
2
1
7
5
1
2
8
4
5
9
7
6
3
6
3
4
2
8
7
9
5
1
5
7
1
6
2
3
8
9
4
meta-llama/llama-3.3-70b-instruct
meta
4
1
5
9
8
2
3
6
7
2
3
7
1
5
4
5
8
6
8
9
6
7
3
6
1
2
4
5
8
3
2
1
1
7
4
9
1
4
9
8
6
5
2
3
7
9
5
4
3
7
8
1
6
2
7
2
1
4
9
3
8
5
6
6
7
8
5
4
9
3
1
2
3
6
2
7
1
8
4
9
5
deepseek/deepseek-chat-v3-0324
deepseek
4
6
2
9
5
8
3
1
7
3
9
7
1
4
2
5
8
6
8
5
1
7
3
6
4
2
9
5
8
6
2
7
1
9
4
3
2
3
9
6
8
4
7
5
1
9
1
4
3
6
7
1
2
5
1
2
8
4
9
5
6
7
3
6
4
3
8
2
7
9
5
1
7
5
9
3
1
6
6
8
4
x-ai/grok-3-mini-beta
x-ai
4
1
6
9
8
5
3
7
2
2
9
7
1
4
3
5
8
6
8
3
5
7
2
6
4
1
9
7
8
2
5
9
1
6
4
3
5
6
9
8
3
4
7
2
1
9
4
1
3
7
2
8
6
5
3
2
8
4
6
9
7
5
1
6
5
4
2
1
8
9
3
7
1
7
3
6
5
4
2
8
9
qwen/qwen3-235b-a22b
qwen
4
6
5
9
1
8
3
7
2
1
3
7
2
4
5
8
9
6
8
9
2
7
3
6
4
5
1
2
8
6
5
9
1
7
4
3
5
1
3
6
8
4
2
9
7
9
7
4
3
2
7
1
6
5
3
2
1
4
6
9
5
8
7
6
5
8
8
7
2
9
3
1
1
7
9
5
1
3
6
2
4
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.