Cheese Dossier · CF-1.3

Claude Opus 4.6

anthropic · Cheese #22· 6 signals · specialty knowledge · Proprietary

CHEESE INDEX
63.5
DEPTH
62.7
BALANCE
70.5
CONSENSUS
70.5
STABILITY
82.6
COVERAGE
85%
VALUE
2.12
I / O
$5 / $25
SPECIALTY
knowledge +24.9
BEST FIT
long_horizon 70.6

深度雷達

codingagentmathknowleinstrumultimmultilreason

實線=本模型 · 虛線=群體中位

芝士軸

coding41.6
agent63.5
horizon71.2

Cheese Index = 可用軸中位數;Balance = 三軸分歧小。

能力細節

swe ★80.8
gpqa ★91.3
hle ★53.1
mcp62.7
coding50.3
agent ★63.5
math58.6
knowledge ★79.8
instruction52.6
multimodal59.5
reasoning67.1

部署建議

中段指數:適合成本敏感路徑。優先看 Pareto/Value;用深度雷達找專長匹配,再與更高指數模型 A/B。

鄰近差分

vs GPT-5.6 Luna

coding-10.0
agent+6.7
horizon+4.0