Configs › Qwen3-32B tok/s: Mac Max 28.4, Strix Halo 6.4
Qwen3-32B tok/s: Mac Max 28.4, Strix Halo 6.4
Measured Qwen3-32B speeds: 28.4 tok/s on Mac Max (MLX-4BIT, mlx); 6.4 tok/s on Strix Halo. 3 configs, each linked to its source, snapshot 2026-10-16.
Snapshot 2026-10-16 · 394 configs on AI boxes and 423 on GPU cards · 123 models. Every number is generated from the tracker snapshot and linked to where it was measured; nothing here is typed by hand.
Quick answers
- How fast is Qwen3-32B on Mac Max?
- 28.4 tok/s single-stream decode (MLX-4BIT, mlx), the fastest of 2 measured configs on Apple M-series Max (MacBook Pro / Mac Studio Max), measured by emihiggins, source last checked 2026-09-29. source
- How fast is Qwen3-32B on Strix Halo?
- 6.4 tok/s single-stream decode (Q8_0, llama.cpp/rocm), the only measured config on AMD Strix Halo (Ryzen AI Max+ 395), measured by lhl, source last checked 2026-10-16. source
Measured configs for Qwen3-32B
| Model | Hardware | Quant | Backend | Decode | Context | Trust | Source |
|---|---|---|---|---|---|---|---|
| Qwen3-32B | Mac Max | MLX-4BIT | mlx | 28.4 tok/s (prefill 731) | 512 | Extracted, source-linked | source |
| Qwen3-32B | Mac Max | Q4_K_M | llama.cpp | 26.4 tok/s (prefill 667) | 512 | Extracted, source-linked | source |
| Qwen3-32B | Strix Halo | Q8_0 | llama.cpp/rocm | 6.4 tok/s (prefill 226.1) | Structured table | source |
More: all configs · hardware · news · contributors · methodology · llms.txt · llms-full.txt · API (OpenAPI) · JSON snapshot. Built by Altronis, private on-prem AI, Singapore.