Recipe
qwen3-6-27b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-mi7oytym
qwen3-6-27b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-mi7oytymObserved LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Engine version
- b8870
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 32,768
- Max concurrency
- 8
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/unsloth/Qwen3.6-27B-GGUF- Repository
- unsloth/Qwen3.6-27B-GGUF
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Candidate: useful compatibility or speed evidence. The registry does not offer Run until promotion requirements are met.
Observed configuration
llama.cpp
Evidence only · candidate · reference
Candidate evidence — not a Run contract
Observed source tokens
Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.
llama-server-mqwen3.6-27b-unsloth-ud-q4kxl.gguf-ngl999-fa1--no-mmap--mlock-c32768-ctkq8_0-ctvq8_0-b4096-ub4096--parallel8
| Flag | Value |
|---|---|
-m | qwen3.6-27b-unsloth-ud-q4kxl.gguf |
-ngl | 999 |
-fa | 1 |
-c | 32768 |
-ctk | q8_0 |
-ctv | q8_0 |
-b | 4096 |
-ub | 4096 |
--parallel | 8 |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 32,768 | — | 11.5 | 1,926.7 | observed | qwen3-6-27b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-mi7oytym-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 1
- hardware id
- ryzen-ai-max-plus-395-128gb
- id
- qwen3-6-27b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-mi7oytym
- model instance id
- unsloth-qwen3-6-27b-gguf--ud-q4-k-xl
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- qwen3-6-27b-ud-q4-k-xl-ryzen-ai-max-plus-395-128gb-llama-cpp-tp1-mi7oytym-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- b8870
serving
- max concurrency
- 8
- max context tokens
- 32,768
- tensor parallel
- 1
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-30T09:10:02Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-30T09:10:02Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-30T09:10:02Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
- serving.max concurrency · provenance · captured at
- 2026-08-31T23:03:15Z
serving.max concurrency · reason server-capacity-derived-from-source-evidence
serving.max concurrency · state known
metadata
- localmaxxing · backend
- vulkan
- localmaxxing · batch size
- 1
- localmaxxing · hardware label
- Ryzen AI Max 395
- localmaxxing · notes
- M5 Strix Halo. Qwen 3.6 27B DENSE at unsloth UD-Q4_K_XL (17 GB). Single-stream 512/512 vllm bench serve, 20 prompts. KV q8_0. Dense 27B is dramatically slower than 35B-A3B MoE on this rig (11.46 vs 52.12 t/s for same Q4 quant) — Strix Halo iGPU has no compute headroom for dense; MoE sparse activation (3B/35B per token) is the only workable path on this hardware.
- localmaxxing · observed command
- llama-server -m qwen3.6-27b-unsloth-ud-q4kxl.gguf -ngl 999 -fa 1 --no-mmap --mlock -c 32768 -ctk q8_0 -ctv q8_0 -b 4096 -ub 4096 --parallel 8
- localmaxxing · run id
- cmofby2jr0001ju04mi7oytym
- localmaxxing · tokenized · arguments
- llama-server, -m, qwen3.6-27b-unsloth-ud-q4kxl.gguf, -ngl, 999, -fa, 1, --no-mmap, --mlock, -c, 32768, -ctk, q8_0, -ctv, q8_0, -b, 4096, -ub, 4096, --parallel, 8
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-recipe | www.localmaxxing.com/en/runs/cmofby2jr0001ju04mi7oytym ↗ |