Verified65% confidenceFactTime unknown
Gemma 4的1B/2B模型量化后仅需4GB显存即可运行
3
Sources
65%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe FP16 version of Gemma 3 12B requires approximately 24 GB of VRAM, while the Q4_K_M version reduces this to approximately 7–8 GB, with inference speed improving by approximately 2–3x and structured output accuracy losing only approximately 2–5%.76% similarUnverifiedThe Gemma 4 26B model is approximately 17 GB and is best suited for GPUs with 24 GB or more of VRAM, delivering the best programming performance76% similarUnverifiedAfter quantization, Gemma 4 31B has a model size of approximately 20GB, with measured GPU memory usage of approximately 20.5GB and GPU utilization of approximately 91%.74% similarUnverifiedThe Gemma 4 26B MOE version achieves an actual inference speed of 60 tokens per second on a GPU with 24GB of VRAM.73% similarUnverifiedThe Gemma 4 26B MOE version can run on a graphics card with 24GB of VRAM.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/38370API
curl https://kongchang.com/api/v1/knowledge/claims/38370MCP
get_claim(id=38370)