Local

Studio

Filters open-weight models with fewer than 40 billion parameters from the Overall ranking, preserves its raw Elo ratings and relative order, then normalizes within this board.Updated 09/10, 21:09

#ModelScore

1

Qwen3.8-27BAlibaba

100.0

2

Qwen3.6-27BAlibaba

89.8

3

Muse Glimmer 30BMeta

88.8

4

Qwen3.6-35B-A3BAlibaba

86.5

5

Qwen3.5-27BAlibaba

85.6

6

Gemma 4 31BGoogle

84.7

7

Qwen3.5-9BAlibaba

82.9

8

Qwen3.5-35B-A3BAlibaba

82.4

9

Qwen3-30B-A3B-Thinking-2507Alibaba