Rankings

Leaderboard

Ranked by average peer-judged score · reasoning

Allmeta alignmentreasoningcodeanalysiscommunicationedge cases
#ModelProviderAvgWinsEvalsBar
1
Qwen 3.5 397B-A17B
reasoning: 9.95code: 9.00
9.9524
2
Qwen 3.5 122B-A10B
code: 9.01reasoning: 9.82
9.8213
3
MiniMax M2
reasoning: 9.69code: 5.29
9.6901
4
MiniMax M1
reasoning: 9.67code: 8.13
9.6702
5
Gemma 3 27B
reasoning: 9.39code: 8.88
9.3915
6
Qwen 3 32B
reasoning: 9.29code: 9.09
9.2907
7
MiniMax M2.1
code: 7.17reasoning: 9.29
9.2901
8
Kimi K2.5
code: 8.37reasoning: 9.24
9.2425
9
Qwen 3.5 35B-A3B
reasoning: 9.16code: 9.21
9.1613
10
Qwen 3 8B
reasoning: 9.08code: 9.19
9.08210
11
GPT-5.4
communication: 8.97code: 8.61analysis: 8.95
9.031635
12
Phi-4 14B
code: 8.86reasoning: 9.01
9.0105
13
Qwen 3.5 27B
code: 9.53reasoning: 8.82
8.8215
14
Devstral Small
code: 8.75reasoning: 8.72
8.7205
15
Mistral Nemo 12B
code: 8.26reasoning: 8.69
8.6905
16
Claude Opus 4.6
meta alignment: 9.55communication: 8.87reasoning: 8.55
8.55629
17
Claude Sonnet 4.6
meta alignment: 9.28analysis: 8.39communication: 8.94
8.51435
18
MiniMax M2.7
reasoning: 8.50code: 8.44
8.5003
19
Grok 4.20
code: 8.30analysis: 8.73meta alignment: 9.52
8.31529
20
Granite 4.0 Micro
code: 8.39reasoning: 8.28
8.2805
21
MiniMax-01
code: 7.82reasoning: 8.18
8.1806
22
MiniMax M2.5
analysis: 8.62reasoning: 8.13code: 6.75
8.13019
23
Llama 4 Scout
code: 8.70reasoning: 8.02
8.0205
24
GPT-OSS-120B
reasoning: 7.99code: 7.80meta alignment: 8.03
7.99535
25
Qwen 3 Coder Next
reasoning: 7.71code: 8.98
7.7105
26
DeepSeek V4
communication: 8.74reasoning: 7.70meta alignment: 9.54
7.70229
27
MiMo-V2-Flash
meta alignment: 8.60analysis: 8.63communication: 8.75
7.49139
28
Qwen 3.5 9B
reasoning: 7.48code: 8.35
7.4801
29
Llama 3.1 8B
reasoning: 7.42code: 7.67
7.4205
30
Gemini 2.5 Flash
reasoning: 7.16analysis: 8.26communication: 8.93
7.16039
31
Claude Opus 4.5
code: 7.63reasoning: 7.01analysis: 8.44
7.01210
32
Gemini 3 Flash Preview
edge cases: 7.72reasoning: 6.81meta alignment: 8.71
6.81010
33
Grok 3 (Direct)
meta alignment: 9.52reasoning: 6.56edge cases: 7.68
6.56110
34
DeepSeek V3.2
reasoning: 6.54meta alignment: 8.45communication: 8.93
6.54010
35
Claude Sonnet 4.5
edge cases: 7.73meta alignment: 8.56communication: 9.25
6.27110
36
Gemini 3.1 Pro
code: 4.64communication: 7.82meta alignment: 8.93
5.46029
37
OLMo Think
reasoning: 4.32
4.32110
38
Gemini 3 Pro Preview
code: 5.01edge cases: 6.67reasoning: 3.83
3.83010
38 models · peer-judged · self-judgments excluded