Released: 2026-05-10
google/gemini-3-flash-preview
6
2
4
9
7
1
5
3
8
7
5
3
6
8
2
9
4
1
8
9
1
3
5
4
7
6
2
1
7
6
8
4
9
2
5
3
9
5
8
2
3
7
1
6
4
4
3
2
5
1
6
9
8
7
2
8
5
7
6
4
3
9
1
1
6
9
4
9
3
8
2
5
5
4
7
2
3
8
6
1
9
google/gemini-2.0-flash-001
3
2
5
9
1
4
6
7
8
9
5
3
6
7
8
1
4
2
8
7
1
3
5
2
4
9
6
5
1
6
8
4
9
2
3
7
6
4
8
2
3
7
1
5
9
4
3
2
1
8
6
5
7
9
2
8
9
7
6
5
3
15
4
7
6
4
1
9
3
8
2
5
1
9
3
5
2
16
7
8
3
openai/gpt-4o
openai
1
2
7
9
3
4
5
6
8
6
5
3
2
8
1
4
7
9
8
9
1
6
5
7
3
2
4
5
7
6
8
1
3
2
9
0
9
1
8
5
2
7
6
4
3
4
3
2
0
9
6
7
8
1
3
8
5
7
4
2
1
5
6
7
6
9
1
6
3
8
2
5
2
4
4
4
7
5
9
1
7
meta-llama/llama-3.3-70b-instruct
meta
5
2
4
9
1
3
7
6
8
1
5
3
2
6
4
9
8
7
8
9
1
5
7
2
6
4
3
3
6
2
8
4
1
5
7
9
9
1
8
3
5
7
2
4
6
4
7
5
6
8
9
1
3
2
2
8
6
1
3
5
4
9
7
7
4
9
4
2
8
3
1
5
6
3
7
7
9
6
8
5
1
anthropic/claude-3.7-sonnet
anthropic
3
2
7
9
1
4
5
6
8
6
5
3
2
7
8
4
9
1
8
4
1
6
5
3
7
2
9
5
7
6
8
3
9
2
4
1
9
6
8
4
2
7
1
5
3
4
3
2
1
9
6
5
8
7
1
8
9
7
6
5
3
4
2
7
9
4
6
1
3
8
2
5
2
5
6
3
8
4
9
1
7
mistralai/mistral-medium-3
mistral
4
2
6
9
3
1
5
7
8
7
5
3
2
8
4
6
9
1
8
9
1
6
5
7
3
4
2
5
7
6
8
1
3
2
4
9
9
4
8
5
2
7
1
6
3
4
3
2
1
9
6
8
5
7
6
8
5
7
4
2
9
3
1
1
6
7
4
9
3
8
2
5
2
3
9
8
6
5
4
1
7
deepseek/deepseek-chat-v3-0324
deepseek
7
2
6
9
3
4
5
1
8
9
5
3
1
6
8
2
4
7
8
4
1
2
5
7
6
3
9
1
9
6
8
4
5
2
7
3
5
8
7
3
2
9
1
6
4
4
3
2
5
1
6
9
8
7
2
1
5
7
8
9
4
3
6
6
7
4
1
9
3
8
2
5
3
6
9
4
7
2
8
1
9
x-ai/grok-3-mini-beta
x-ai
1
2
4
9
3
5
6
7
8
6
5
3
1
7
8
9
4
2
8
7
1
2
5
4
3
6
9
7
9
6
8
4
1
2
5
3
3
4
8
6
2
7
5
9
1
4
3
2
5
9
6
7
8
1
2
6
5
7
8
9
1
3
4
9
1
7
4
6
3
8
2
5
5
8
9
3
1
2
4
6
7
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.