Models
Every model in the arena, ranked by blind human and model judgment.
All models
GPT-6.1 Terra
#1 · 1113 ±20
GPT-6.1 Sol
#2 · 1107 ±17
Claude Fable 5.2
#3 · 1106 ±18
Hy4
#4 · 1087 ±22
Muse Wonder 1.0
#5 · 1059 ±21
Kimi K3.1 Pro
#6 · 1047 ±21
Gemini 4 Argon
#7 · 1047 ±21
GPT-6.2 Sol
#8 · 1041 ±60
Claude Opus 5.5
#9 · 1041 ±19
Gemini 4 Carbon
#10 · 1035 ±22
GPT-6 Astra
#11 · 1030 ±20
GPT-6.2 Astra
#12 · 1024 ±54
Claude Fable 5
#13 · 1021 ±21
Claude Fable 5.1
#14 · 1019 ±18
Qwen3.8-2.4T-A95B
#15 · 992 ±21
Grok 4.7
#16 · 990 ±21
Muse Spark 1.3
#17 · 986 ±21
Kimi K3
#18 · 984 ±20
MiniMax M3.1 Flash
#19 · 978 ±20
GPT-6 Luna
#20 · 973 ±20
Gemini 3.8 Flash
#21 · 961 ±21
Claude Sonnet 5.5
#22 · 955 ±22
MiMo V2.6 Pro
#23 · 952 ±21
DeepSeek V4.1 Flash 1017
#24 · 939 ±25
Claude Haiku 5.5
#25 · 938 ±21
GLM 5.3 Flash
#26 · 926 ±22
Computer board rank, then rating and its 95% interval, drawn on one scale with lines at 1000 and 1100. n = 26 models, each rated from at least 30 comparisons. Last updated .