Configs › Qwen3.8-27B VL on Strix Halo: 22 tok/s measured
Qwen3.8-27B VL on Strix Halo: 22 tok/s measured
Measured Qwen3.8-27B VL speeds: 22 tok/s on Strix Halo. 1 config, each linked to its source, snapshot 2026-10-30.
Snapshot 2026-10-30 · 412 configs on AI boxes and 288 on GPU cards · 121 models. Every number is generated from the tracker snapshot and linked to where it was measured; nothing here is typed by hand.
Fastest measured single-stream run of Qwen3.8-27B VL: 22 tok/s on Strix Halo (UD-Q4_K_XL, llama.cpp, speculative decode, mtp), snapshot 2026-10-30.
Fastest measured run on each box
| Hardware | Decode tok/s | Quant | Backend | Context | Trust | Configs on this box | Source |
|---|---|---|---|---|---|---|---|
| Strix Halo | 22 | UD-Q4_K_XL | llama.cpp (speculative decode, mtp) | Owner-submitted | 1 | source |
Questions and answers
- How fast is Qwen3.8-27B VL on local hardware?
- The fastest measured single-stream run is 22 tok/s on Strix Halo (UD-Q4_K_XL, llama.cpp, speculative decode, mtp). TokenMark tracks 1 single-stream config for Qwen3.8-27B VL on 1 box.
- How many measured configs does TokenMark have for Qwen3.8-27B VL?
- 1 single-stream config on 1 box, newest source check 2026-09-16. Every row links to where it was measured.
All measured configs for Qwen3.8-27B VL
| Model | Hardware | Quant | Backend | Decode | Context | Trust | Source |
|---|---|---|---|---|---|---|---|
| Qwen3.8-27B VL | Strix Halo | UD-Q4_K_XL | llama.cpp (speculative decode, mtp) | 22 tok/s | Owner-submitted | source |
Last updated: snapshot 2026-10-30; newest source check 2026-09-16. Rankings use single-stream runs only; aggregate and multi-GPU rows are kept off them.
More: all configs · hardware · news · contributors · methodology · llms.txt · llms-full.txt · API (OpenAPI) · JSON snapshot. Built by Altronis, private on-prem AI, Singapore.