Qwen3.6-35B-A3B Q4_K_M on Ryzen AI Max+ 395 (llama.cpp Vulkan) — inter-token latency, 4 concurrent requests: 30.0 ms/token

The measurement

On a Beelink GTR9 Pro — AMD Ryzen AI Max+ 395, Radeon 8060S (gfx1151), 128 GB LPDDR5X-8000 unified running llama.cpp (server, Vulkan backend), b9049 (server_fingerprint b9049-2496f9c14) with ggml-org/Qwen3.6-35B-A3B-GGUF @ baec3ebee244 (Q4_K_M), the measured inter-token latency, 4 concurrent requests was 30.0 ms/token (median over valid requests). The measurement was taken under the frozen endurance-30m-v1@2026-08-03 workload; evidence level lab_single_run, 2 valid runs across 2 physical units. The value is re-derived from the raw run records on every CI build.

Statement

Median inter-token latency sustained over the full 3-hour pass at closed-loop concurrency 4, both units pooled.

Evidence

System

Beelink GTR9 Pro — AMD Ryzen AI Max+ 395, Radeon 8060S (gfx1151), 128 GB LPDDR5X-8000 unified

Runtime

llama.cpp (server, Vulkan backend), b9049 (server_fingerprint b9049-2496f9c14)

Model artifact

ggml-org/Qwen3.6-35B-A3B-GGUF @ baec3ebee244 (Q4_K_M)

Scope

endurance-30m-v1@2026-08-03

Aggregation

median over valid requests

Evidence level

lab_single_run

Status

active

Limitations

One 3-hour pass per unit — the pattern reproduced on both commercially identical units, but repeats within a unit are pending. Closed-loop concurrency 4, reasoning disabled, this corpus — not a general thermals verdict. Ambient not instrumented; die temperatures documented in the run notes.

Runs

• run-20260803-end-c4-a
• run-20260803-end-c4-b

Derivation

The value is re-derived from the raw run records by this query on every CI build — a published number cannot silently drift from its evidence.

catalog/claims/sql/itl-median.sql

Cite this claim

AGmind Systems Lab. "Median inter-token latency sustained over the full 3-hour pass at closed-loop concurrency 4, both units pooled." Claim strix.qwen36.endurance.c4.itl-median (30.0 ms/token), evidence level lab_single_run, scope endurance-30m-v1@2026-08-03. https://agmind.ai/claims/strix.qwen36.endurance.c4.itl-median/

@misc{agmind_strix_qwen36_endurance_c4_itl_median,
  author       = {{AGmind Systems Lab}},
  title        = {Median inter-token latency sustained over the full 3-hour pass at closed-loop concurrency 4, both units pooled.},
  howpublished = {\url{https://agmind.ai/claims/strix.qwen36.endurance.c4.itl-median/}},
  note         = {Claim strix.qwen36.endurance.c4.itl-median: 30.0 ms/token; evidence level lab\_single\_run; scope endurance-30m-v1@2026-08-03},
  year         = {2026}
}

Machine-readable: /claims/strix.qwen36.endurance.c4.itl-median.json · full registry