LLMCheck: Apple M4 Max, MLX, Qwen3.6-35B-A3B Q4_K_M decode
48.0 tok/s (decode, per-request)
Fields
id
benchmarks/llmcheck-m4-max-qwen36-35b-a3b-mlx-decodeMetric
decode_tok_s
Value
48
Unit
tok/s (decode, per-request)
Model
Qwen3.6-35B-A3B
Methodology
THIRD-PARTY MEASURED (LLMCheck open dataset). M4 Max, MLX, Qwen3.6-35B-A3B Q4_K_M, per-request decode.
Measured by
third_party
Reproducible
yes
notes
Gives apple-m4-max a MoE data point on the target model class; ~13% slower than M5 Max on the same model/quant, consistent with 546 vs 614 GB/s bandwidth.
Notes
Gives apple-m4-max a MoE data point on the target model class; ~13% slower than M5 Max on the same model/quant, consistent with 546 vs 614 GB/s bandwidth.