Unverified50% confidenceBenchmarkExact time
27B稠密模型在24GB M4 Pro Mac mini上使用4-bit量化可以运行，速度约为11 tok/s
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/20/2026
First Seen
Valid until: 11/18/2026
Sources
Related Claims
UnverifiedOn a 128GB MacBook Pro M3 Max, independent benchmarks show DS4 achieving a generation speed of 26.68 tokens per second.79% similarUnverifiedOn a 16 GB M4 Mac Mini, after loading Ornith 9B GGUF, 6.2 GB is used for model weights, and the maximum context length can reach 260,000 tokens.78% similarUnverifiedThe test was conducted on a MacBook with 36GB of memory, with the model loaded at a 128K context size and Q8 KV cache.78% similarUnverifiedAfter 4-bit quantization, Llama 3.3 70B has a model size of approximately 40 GB and can run on a Mac Studio with 64 GB of memory or a single professional GPU with 48 GB of VRAM.78% similarUnverifiedDense models at the 30B scale are considered the sweet spot for local deployment, capable of running on Apple Silicon devices such as M2 Ultra (192GB) and M4 Pro (48GB), or on an RTX 4090 (24GB VRAM) with the aid of quantization techniques.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/779028API
curl https://kongchang.com/api/v1/knowledge/claims/779028MCP
get_claim(id=779028)