Local AI Registry

Recipe

qwen3-6-27b-q4-k-xl-rtx-3090-24gb-llama-cpp-tp1-k1vkl3qa

qwen3-6-27b-q4-k-xl-rtx-3090-24gb-llama-cpp-tp1-k1vkl3qa
JSON APIFind compatible recipes

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

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Accelerators
1
Tensor parallel
1
Context tokens
2,048
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/AtomicChat/Qwen3.6-27B-UDT-MTP-GGUF
Repository
AtomicChat/Qwen3.6-27B-UDT-MTP-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

Source
https://www.localmaxxing.com/en/runs/cmpsnmdz9009zmq01k1vkl3qa

Observed source tokens

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

  1. -m
  2. /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf
  3. -md
  4. /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf
  5. #
  6. --mmproj
  7. /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/mmproj-BF16.gguf
  8. --host
  9. 127.0.0.1
  10. --port
  11. 18080
  12. -ngl
  13. 99
  14. -ngld
  15. 99
  16. -fa
  17. on
  18. -c
  19. 150000
  20. --parallel
  21. 1
  22. --spec-type
  23. nextn
  24. --draft-max
  25. 2
  26. --draft-min
  27. 1
  28. -ctk
  29. turbo3
  30. -ctv
  31. turbo3
  32. --cache-ram
  33. 3000
  34. --jinja
  35. --temp
  36. 0.85
  37. --top-k
  38. 20
  39. --top-p
  40. 0.95
  41. --min-p
  42. 0.0
  43. --presence-penalty
  44. 1.5
  45. --repeat-penalty
  46. 1.0
FlagValue
-m/home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf
-md/home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf
--mmproj/home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/mmproj-BF16.gguf
--host127.0.0.1
--port18080
-ngl99
-ngld99
-faon
-c150000
--parallel1
--spec-typenextn
--draft-max2
--draft-min1
-ctkturbo3
-ctvturbo3
--cache-ram3000
--temp0.85
--top-k20
--top-p0.95
--min-p0.0
--presence-penalty1.5
--repeat-penalty1.0

Environment

VariableValue
ExecStart/home/garrett/llama.cpp/build/bin/llama-server

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
12,048—56.2132observedqwen3-6-27b-q4-k-xl-rtx-3090-24gb-llama-cpp-tp1-k1vkl3qa-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
rtx-3090-24gb
id
qwen3-6-27b-q4-k-xl-rtx-3090-24gb-llama-cpp-tp1-k1vkl3qa
model instance id
atomicchat-qwen3-6-27b-udt-mtp-gguf--q4-k-xl
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-27b-q4-k-xl-rtx-3090-24gb-llama-cpp-tp1-k1vkl3qa-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, AtomicChat/Qwen3.6-27B-UDT-MTP-GGUF:Q4_K_XL, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 2048
container port
8,080
environment · LLAMA CACHE
/root/.cache/huggingface
host port
8,080
image
ghcr.io/ggml-org/llama.cpp:server-cuda12-b10481@sha256:b2497f8834f5ecb4e38530f6bf2734b8e0be107f0f0857e259672d1cb85b71c2
ipc
host
kind
docker
shm size
16g
synthesized · generated at
2026-08-31T22:12:17Z
synthesized · image provenance
gemma-4-12b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1
synthesized · template
llama-cpp-server-v1

mounts

read onlytarget
No/root/.cache/huggingface

engine

name
llama.cpp

serving

max concurrency
1
max context tokens
2,048
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

metadata

localmaxxing · hardware label
RTX 3090
localmaxxing · notes
Using Atomic Chat fork of llama.cpp with Turboquant and spec decoding of 2. 3090 running at 480w pwr limit.
localmaxxing · observed command
ExecStart=/home/garrett/llama.cpp/build/bin/llama-server \ -m /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf \ -md /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf \ # --mmproj /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/mmproj-BF16.gguf \ --host 127.0.0.1 \ --port 18080 \ -ngl 99 \ -ngld 99 \ -fa on \ -c 150000 \ --parallel 1 \ --spec-type nextn \ --draft-max 2 \ --draft-min 1 \ -ctk turbo3 \ -ctv turbo3 \ --cache-ram 3000 \ --jinja \ --temp 0.85 \ --top-k 20 \ --top-p 0.95 \ --min-p 0.0 \ --presence-penalty 1.5 \ --repeat-penalty 1.0
localmaxxing · run id
cmpsnmdz9009zmq01k1vkl3qa
localmaxxing · tokenized · arguments
-m, /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf, -md, /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/Qwen3.6-27B-UDT-Q4_K_XL_MTP.gguf, #, --mmproj, /home/garrett/.cache/huggingface/hub/models--AtomicChat--Qwen3.6-27B-UDT-MTP-GGUF/snapshots/4b31c9f1e5be79eaea778e24ac50aab8a19e6dad/mmproj-BF16.gguf, --host, 127.0.0.1, --port, 18080, -ngl, 99, -ngld, 99, -fa, on, -c, 150000, --parallel, 1, --spec-type, nextn, --draft-max, 2, --draft-min, 1, -ctk, turbo3, -ctv, turbo3, --cache-ram, 3000, --jinja, --temp, 0.85, --top-k, 20, --top-p, 0.95, --min-p, 0.0, --presence-penalty, 1.5, --repeat-penalty, 1.0
localmaxxing · tokenized · environment · ExecStart
/home/garrett/llama.cpp/build/bin/llama-server
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/cmpsnmdz9009zmq01k1vkl3qa ↗