Expired50% confidenceBenchmarkExact time
In Ollama's CPU inference mode, generation speed drops from 30–80 tokens per second on GPU to 2–8 tokens per second.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/9/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 10/7/2026(expired)
Sources
Complete Guide to Deploying Dify Locally: Docker + Ollama + RAG Setup Tutorial
bilibili前端vue实战项目7/6/2026
Related Claims
UnverifiedCPU推理速度通常为GPU速度的1/5到1/1075% similarUnverified配备NVIDIA GPU并使用CUDA加速，7B模型可以达到50-100+ token/s的推理速度73% similarUnverifiedCPU+GPU hybrid inference uses a layer-level offloading mechanism, controlling the number of offloaded layers via the n_gpu_layers parameter; the main bottleneck is PCIe bandwidth latency, which typically reduces generation speed to 1/3 to 1/2 of pure GPU inference.71% similarUnverifiedA mid-range GPU can achieve tens to hundreds of times the matrix computation throughput of a CPU70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/374055API
curl https://kongchang.com/api/v1/knowledge/claims/374055MCP
get_claim(id=374055)