Cheese Dossier · CF-1.3
Gemini 3.1 Flash-Lite
google · Cheese #168· 4 signals · specialty gpqa · Proprietary
CHEESE INDEX
40.5
DEPTH
40.5
BALANCE
74.6
CONSENSUS
61.4
STABILITY
99.6
COVERAGE
46%
VALUE
23.17
I / O
$0.25 / $1.5
SPECIALTY
gpqa +11.9
BEST FIT
budget 36.5
深度雷達
實線＝本模型 · 虛線＝群體中位
芝士軸
coding
agent
horizon
Cheese Index = 可用軸中位數；Balance = 三軸分歧小。
能力細節
gpqa ★
hle
coding
agent
knowledge
multimodal
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-5.2
coding+2.8
agent+20.0
horizon-28.5
指數鄰近
- GPT-5.2 40.7
- Gemini 2.5 Pro 40.4
- Step 3.5 Flash 40.8
- MiMo-V2-Omni 40.2
- GPT-5 mini 41.0
同組織
- Gemini 3.5 Flash 68.1
- Gemini 3 Flash 52.2
- Gemini 3.6 Flash 51.9
- Gemma 4 26B A4B 47.5