Skip to content
VECTOR WIREAI INTELLIGENCE
UTC

CMMLU leaderboard — the family board.

3 TRACKED VARIANTS · 1 FIRST-CLASS BOARD

One family, 3 tracked prints. Scoring configs and splits of the base board render as sections of this page, each under its own honest label.

● LIVE DATA27 MODELS ON THIS BOARDLEADER KIMI K2 BASE AT 90.9

The board — CMMLUTOP 27 OF 27 MEASURED · BEST SCORE PER MODEL, DEFAULT VARIANT

#ModelVendorScoreVLast seen
01Kimi K2 BaseMoonshot90.92026-06-13
02DeepSeek-V4-Pro-BaseDeepSeek90.82026-06-05
03DeepSeek-V4-Flash-BaseDeepSeek90.42026-06-05
04Hunyuan Large InstructTencent90.42026-05-31
05MiMo V2.5 ProXiaomi90.22026-08-23
06MiMo V2.5 Pro BaseXiaomi90.22026-06-05
07Hunyuan LargeTencent90.22026-05-31
08Qwen2 72BAlibaba90.12026-09-10
09Qwen2 Instruct (72B)Alibaba90.12026-08-23
10Hy3-preview-BaseTencent89.62026-06-05
11Qwen2.5 Instruct 72BAlibaba89.52026-05-01
12DeepSeek-V3.2-Exp-BaseDeepSeek88.92026-06-13
13Step3.5 Flash BaseStepFun88.92026-06-05
14DeepSeek-V3.1-BaseDeepSeek88.82026-06-13
15DeepSeek-V3DeepSeek88.82026-05-01
16MiMo V2.5 BaseXiaomi88.22026-06-05
17DeepSeek-V3-BaseDeepSeek87.92026-06-05
18MiMo V2 Flash BaseXiaomi87.42026-06-13
19GLM-4.5-BaseZ.ai86.52026-06-05
20DeepSeek-V2DeepSeek842026-09-10
21LongCat Flash LiteMeituan82.52026-08-23
22MoonlightMoonshot78.22026-06-12
23Qwen2.5 3BAlibaba752026-06-05
24Llama 3.1 Instruct 405BMeta73.72026-05-01
25Llama 3 (70B)Meta67.22026-09-10
26Llama 3 Instruct 8BMeta50.82026-08-26
27ERNIE 4.5Baidu39.82026-08-23
◆ IV — INDEPENDENTLY VERIFIED · ◆ AA — AGGREGATOR · ◆ VA — VENDOR · ◇ SA — SOURCE-ATTRIBUTED

Variants of this boardSCORING CONFIGS & SPLITS — NOT CITED STANDALONE, SO THEY LIVE HERE

Other tracked prints

THIN BOARDSunder 5 published models each — listed, never hidden
PrintLeaderScoreModels
CMMLU (Acc.)Qwen2.5 Instruct 72B89.54
CMMLU (EM)DeepSeek-V4-Pro-Base90.83
THE FAMILY BOARD · EVERY PRINT LABELED · EVERY SCORE SOURCE-ATTRIBUTED ON ITS OWN LIVE BOARDAll leaderboardsThe frontier boardModels by vendor