# Qwen3.6-35B-A3B Q4_K_M on Ryzen AI Max+ 395 (llama.cpp Vulkan) — strict-JSON task success: 100.0 % of requests

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 strict-JSON task success was 100.0 % of requests (share of all issued requests). The measurement was taken under the frozen structured-agent-v1@2026-08-03 workload; evidence level lab_repeated, 3 valid runs across 1 physical unit. The value is re-derived from the raw run records on every CI build.

## Statement

> Share of strict-JSON automation requests whose output parsed and matched the ground truth per key, reasoning disabled.

## Evidence

| Field | Value |
| --- | --- |
| Claim id | `strix.qwen36.structured.c1.task-success-nothink` |
| Value | 100.0 % of requests |
| Metric | strict-JSON task success |
| Aggregation | share of all issued requests |
| 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) |
| Workload scope | structured-agent-v1@2026-08-03 |
| Evidence level | lab_repeated |
| Status | active |
| Runs | 3 across 1 unit(s) |
| Published | 2026-08-03 |
| Last changed | 2026-08-03 |

**Limitations.** Three repeated runs on one unit. Deterministic short tasks with closed label sets — no tool execution, no sandbox. Concurrency 1 only.

## Raw evidence

- Sealed run bundle `run-20260803-sa-nothink-a`: https://github.com/botAGI/agmind-lab/tree/main/runs/run-20260803-sa-nothink-a
- Sealed run bundle `run-20260803-sa-nothink-a-r2`: https://github.com/botAGI/agmind-lab/tree/main/runs/run-20260803-sa-nothink-a-r2
- Sealed run bundle `run-20260803-sa-nothink-a-r3`: https://github.com/botAGI/agmind-lab/tree/main/runs/run-20260803-sa-nothink-a-r3
- Derivation query: https://github.com/botAGI/agmind-lab/blob/main/catalog/claims/sql/json-task-success.sql
- Machine-readable: https://agmind.ai/claims/strix.qwen36.structured.c1.task-success-nothink.json · full registry https://agmind.ai/claims.json

## Cited on

- https://agmind.ai/reports/beelink-gtr9-pro-local-llm/
- https://agmind.ai/reports/local-llm-json-reliability/
- https://agmind.ai/reports/local-llm-homelab-hardware/
- https://agmind.ai/reports/q8-0-vs-q4-k-m-strix-halo/
- https://agmind.ai/reports/should-you-disable-thinking-local-llm/
- https://agmind.ai/tested-configurations/strix-halo-qwen36-llamacpp-vulkan-structured/
- https://agmind.ai/compare/llamacpp-vulkan-vs-rocm-strix-halo/
- https://agmind.ai/compare/q4km-vs-q80-qwen36-strix-halo/
- https://agmind.ai/compare/thinking-vs-no-thinking-qwen36/
- https://agmind.ai/measured/qwen3-6-35b-a3b-q4-k-m-on-ryzen-ai-max-395/

## Cite this

AGmind Systems Lab. Qwen3.6-35B-A3B Q4_K_M on Ryzen AI Max+ 395 (llama.cpp Vulkan): strict-JSON task success — 100.0 % of requests (share of all issued requests; 3 runs on 1 unit; evidence level lab_repeated; workload structured-agent-v1@2026-08-03). Claim strix.qwen36.structured.c1.task-success-nothink. https://agmind.ai/claims/strix.qwen36.structured.c1.task-success-nothink/

Registry changes feed: https://agmind.ai/claims/changes.xml · Corrections to published results: https://agmind.ai/errata/ · Data licensed CC BY 4.0.
Source page: https://agmind.ai/claims/strix.qwen36.structured.c1.task-success-nothink/