Released: 2026-04-09
google/gemini-3-flash-preview
2
3
6
4
8
1
7
5
9
8
4
9
6
5
7
1
9
3
1
5
7
3
2
9
4
6
8
6
8
2
7
4
3
5
1
9
4
7
5
1
9
8
3
6
2
3
9
1
5
2
6
8
4
7
7
2
8
9
1
5
6
3
4
5
1
4
2
6
3
9
7
8
9
6
3
8
7
4
2
5
1
mistralai/mistral-medium-3
mistral
2
3
6
4
7
8
4
1
5
4
7
8
6
5
1
2
9
3
1
5
9
3
2
9
7
5
8
7
8
2
9
3
5
5
8
4
9
1
5
8
4
2
3
6
7
3
4
9
7
1
6
9
4
2
5
2
8
5
1
3
6
2
9
6
9
3
2
8
7
1
7
1
8
6
7
1
9
4
8
3
6
google/gemini-2.0-flash-001
9
3
6
4
7
2
1
8
5
8
2
4
1
5
3
6
9
7
1
5
7
6
8
9
4
3
2
6
4
2
7
9
1
5
8
3
7
8
5
9
4
2
3
6
1
3
9
1
8
6
5
7
4
2
4
7
8
3
1
5
2
5
6
5
1
3
2
9
6
8
7
4
2
6
9
5
3
4
8
1
3
openai/gpt-4o
openai
7
3
6
4
8
2
1
5
9
2
8
4
1
5
7
6
9
3
1
5
9
6
3
9
7
2
8
6
1
2
7
9
3
5
8
4
8
9
5
5
2
1
3
6
7
3
7
9
8
2
6
2
4
5
4
2
8
9
1
5
9
3
6
5
4
3
2
6
8
4
7
1
9
6
7
3
7
4
8
1
2
meta-llama/llama-3.3-70b-instruct
meta
5
3
6
4
7
8
9
1
2
7
9
2
5
5
3
4
9
8
1
4
8
9
2
9
6
5
7
9
1
2
7
4
5
3
6
8
4
5
5
1
3
2
7
9
1
3
7
9
6
1
4
2
8
5
2
8
1
3
9
6
5
7
4
8
6
4
2
8
7
1
3
9
6
2
7
8
5
1
4
2
3
anthropic/claude-3.7-sonnet
anthropic
8
3
6
4
7
1
2
5
9
7
4
1
6
5
2
8
9
3
1
5
9
3
6
9
7
2
8
9
1
2
7
8
3
5
1
4
4
8
5
1
9
7
3
6
2
3
7
9
5
2
6
1
4
8
5
9
8
6
1
7
4
3
2
6
4
3
2
8
5
9
7
1
2
6
7
9
3
4
8
1
5
deepseek/deepseek-chat-v3-0324
deepseek
9
3
6
4
8
5
1
2
7
2
8
4
6
5
1
7
9
3
1
5
7
3
2
9
6
4
8
6
4
2
1
9
7
5
8
3
8
1
5
9
4
2
3
6
7
3
7
9
8
6
5
2
4
1
7
2
8
5
1
3
4
6
9
5
9
3
2
6
8
4
7
1
4
6
1
7
3
9
8
5
2
x-ai/grok-3-mini-beta
x-ai
2
3
6
4
7
1
8
5
9
4
1
7
8
5
3
6
9
2
1
5
4
6
2
9
7
3
8
6
8
2
7
4
5
9
1
3
9
7
5
1
8
2
3
6
4
3
4
1
9
6
8
2
7
5
5
9
8
3
1
7
4
2
6
7
6
3
2
9
4
1
8
5
8
5
9
6
3
7
2
4
1
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.