Apple M1 Ultra 64GB Llama-3.3-70B 4bit token-gen
12.6 tok/s 4bit @4K ctx (LM Studio)
Fields
id
benchmarks/comm-apple-m1-ultra-llama33-70b-4bit-tgMetric
decode_tok_s
Value
12.6
Unit
tok/s 4bit @4K ctx (LM Studio)
Model
Llama-3.3-70B
Methodology
Silicon Score Llama-3.3-70B model page row; LM Studio 4bit @4k ctx; ~800 GB/s unified
Measured by
third_party
Thermal state
unknown
notes
Cheapest entry to usable 70B. Canonical llama.cpp M-series table: M1 Ultra 83.7 t/s Q4_0 on 7B-class.
Notes
Cheapest entry to usable 70B. Canonical llama.cpp M-series table: M1 Ultra 83.7 t/s Q4_0 on 7B-class.