Local AI Registry

Recipe

qwen3-6-27b-q4-0-intel-arc-pro-b70-32gb-llama-cpp-tp1

qwen3-6-27b-q4-0-intel-arc-pro-b70-32gb-llama-cpp-tp1
JSON APIFind compatible recipes

Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Engine version
9976 (e3546c794) local SYCL Xe2 M6 BMG AOT build with fused Q6_K draft LM-head top-1
Accelerators
1
Tensor parallel
1
Context tokens
4,096
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/Qwen/Qwen3.6-27B
Repository
Qwen/Qwen3.6-27B
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

Source
https://www.localmaxxing.com/en/runs/cmrjbx8bc02g8mj01yzz2v701

Observed source tokens

Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.

  1. llama-server
  2. -m
  3. Qwen3.6-27B-Q4_0.gguf
  4. -dev
  5. SYCL0
  6. -ngl
  7. 99
  8. -c
  9. 4096
  10. -np
  11. 1
  12. -b
  13. 1024
  14. -ub
  15. 256
  16. -ctk
  17. q8_0
  18. -ctv
  19. q8_0
  20. -fa
  21. on
  22. --spec-type
  23. draft-dflash
  24. --spec-draft-model
  25. Qwen3.6-27B-DFlash-Q8_0.gguf
  26. --spec-draft-type-k
  27. f16
  28. --spec-draft-type-v
  29. f16
  30. --spec-draft-n-max
  31. 5
  32. --cache-ram
  33. 0
  34. --ctx-checkpoints
  35. 0
  36. -fit
  37. off
FlagValue
-mQwen3.6-27B-Q4_0.gguf
-devSYCL0
-ngl99
-c4096
-np1
-b1024
-ub256
-ctkq8_0
-ctvq8_0
-faon
--spec-typedraft-dflash
--spec-draft-modelQwen3.6-27B-DFlash-Q8_0.gguf
--spec-draft-type-kf16
--spec-draft-type-vf16
--spec-draft-n-max5
--cache-ram0
--ctx-checkpoints0
-fitoff

Environment

VariableValue
GGML_SYCL_ENABLE_GRAPH0
GGML_SYCL_FUSE_GDN_CACHE1
GGML_SYCL_XE2_Q4_M6_FFN1
GGML_SYCL_XE2_Q4_M6_GATE_UP1
GGML_SYCL_XE2_Q4_M6_PACK_LIMIT187
GGML_SYCL_XE2_Q6_M6_TOP11
LLAMA_DFLASH_FUSED_TOP11
ONEAPI_DEVICE_SELECTORlevel_zero:*
ZE_AFFINITY_MASK2

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
14,096—47.81,156.5observedqwen3-6-27b-q4-0-intel-arc-pro-b70-32gb-llama-cpp-tp1-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
intel-arc-pro-b70-32gb
id
qwen3-6-27b-q4-0-intel-arc-pro-b70-32gb-llama-cpp-tp1
model instance id
qwen-qwen3-6-27b--q4-0
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-27b-q4-0-intel-arc-pro-b70-32gb-llama-cpp-tp1-sweep
status
candidate

capabilities

engine

name
llama.cpp
version
9976 (e3546c794) local SYCL Xe2 M6 BMG AOT build with fused Q6_K draft LM-head top-1

serving

max concurrency
1
max context tokens
4,096
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
xpu
localmaxxing · batch size
1
localmaxxing · hardware label
Intel Arc Pro B70
localmaxxing · notes
One active generation on one Intel Arc Pro B70. The fixed realistic cold suite passed with cached_tokens=0 for all 12 prompts. Native DFlash accepted tokens are verified by the unchanged Q4_0 target. This row adds a fused, exact Q6_K x Q8_1 M=6 draft LM-head top-1 boundary to the promoted 187-projection Xe2 verifier and GDN snapshot-cache fusion. The exact matching AOT control reproduced 44.2205 tok/s, making this confirmed 47.8188 tok/s result a measured 8.14% end-to-end fusion gain and an 8.05% improvement over the prior 44.2554 tok/s LocalMaxxing record. The first independent strict candidate also passed at 47.1143 tok/s. The fused path fails closed and re-decodes ordinary logits after compact-read failure.
localmaxxing · observed command
ONEAPI_DEVICE_SELECTOR=level_zero:* ZE_AFFINITY_MASK=2 GGML_SYCL_ENABLE_GRAPH=0 GGML_SYCL_XE2_Q4_M6_FFN=1 GGML_SYCL_XE2_Q4_M6_PACK_LIMIT=187 GGML_SYCL_XE2_Q4_M6_GATE_UP=1 GGML_SYCL_FUSE_GDN_CACHE=1 LLAMA_DFLASH_FUSED_TOP1=1 GGML_SYCL_XE2_Q6_M6_TOP1=1 llama-server -m Qwen3.6-27B-Q4_0.gguf -dev SYCL0 -ngl 99 -c 4096 -np 1 -b 1024 -ub 256 -ctk q8_0 -ctv q8_0 -fa on --spec-type draft-dflash --spec-draft-model Qwen3.6-27B-DFlash-Q8_0.gguf --spec-draft-type-k f16 --spec-draft-type-v f16 --spec-draft-n-max 5 --cache-ram 0 --ctx-checkpoints 0 -fit off
localmaxxing · run id
cmrjbx8bc02g8mj01yzz2v701
localmaxxing · tokenized · arguments
llama-server, -m, Qwen3.6-27B-Q4_0.gguf, -dev, SYCL0, -ngl, 99, -c, 4096, -np, 1, -b, 1024, -ub, 256, -ctk, q8_0, -ctv, q8_0, -fa, on, --spec-type, draft-dflash, --spec-draft-model, Qwen3.6-27B-DFlash-Q8_0.gguf, --spec-draft-type-k, f16, --spec-draft-type-v, f16, --spec-draft-n-max, 5, --cache-ram, 0, --ctx-checkpoints, 0, -fit, off
localmaxxing · tokenized · environment · GGML SYCL ENABLE GRAPH
0
localmaxxing · tokenized · environment · GGML SYCL FUSE GDN CACHE
1
localmaxxing · tokenized · environment · GGML SYCL XE2 Q4 M6 FFN
1
localmaxxing · tokenized · environment · GGML SYCL XE2 Q4 M6 GATE UP
1
localmaxxing · tokenized · environment · GGML SYCL XE2 Q4 M6 PACK LIMIT
187
localmaxxing · tokenized · environment · GGML SYCL XE2 Q6 M6 TOP1
1
localmaxxing · tokenized · environment · LLAMA DFLASH FUSED TOP1
1
localmaxxing · tokenized · environment · ONEAPI DEVICE SELECTOR
level_zero:*
localmaxxing · tokenized · environment · ZE AFFINITY MASK
2
localmaxxing · tokenized · fidelity
faithful

provenance

captured at
2026-08-30T09:10:02Z

sources

captured atkindurl
2026-08-30T09:10:02Znormalized-recipewww.localmaxxing.com/en/runs/cmrjbx8bc02g8mj01yzz2v701 ↗