Recipe

qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-jm347jzo

qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-jm347jzo

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

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Engine version
b1540
Accelerators
1
Tensor parallel
1
Context tokens
73,728
Max concurrency
1
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

Source
https://www.localmaxxing.com/en/runs/cmoerh82s0001l404jm347jzo

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. gary-qwen36-27b-v4-Q4_K_M.gguf
  4. -ngl
  5. 99
  6. -c
  7. 73728
  8. -np
  9. 1
  10. --cache-type-k
  11. bf16
  12. --cache-type-v
  13. bf16
  14. --jinja
  15. --flash-attn
  16. on
  17. --batch-size
  18. 2048
  19. --ubatch-size
  20. 512
  21. --threads
  22. 8
  23. --threads-batch
  24. 16
  25. --temp
  26. 0.1
  27. --reasoning-format
  28. deepseek
  29. --reasoning-budget
  30. 1024
  31. --no-context-shift
  32. --ctx-checkpoints
  33. 0
  34. --checkpoint-every-n-tokens
  35. -1
FlagValue
-mgary-qwen36-27b-v4-Q4_K_M.gguf
-ngl99
-c73728
-np1
--cache-type-kbf16
--cache-type-vbf16
--flash-attnon
--batch-size2048
--ubatch-size512
--threads8
--threads-batch16
--temp0.1
--reasoning-formatdeepseek
--reasoning-budget1024
--ctx-checkpoints0

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
173,72836.4374observedqwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-jm347jzo-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-ti-24gb
id
qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-jm347jzo
model instance id
unsloth-qwen3-6-27b-gguf--q4-k-m
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-27b-q4-k-m-rtx-3090-ti-24gb-llama-cpp-tp1-jm347jzo-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, unsloth/Qwen3.6-27B-GGUF:Q4_K_M, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 73728
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
version
b1540

serving

max concurrency
1
max context tokens
73,728
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 Ti
localmaxxing · notes
Short-reply variant of cmoer7qzw000ale04gs1qkc1n. Same config, 512 prompt → 128 output. Quick-dispatch tool-call profile. See main entry cmoer7qzw000ale04gs1qkc1n for full DeltaNet stability config.
localmaxxing · observed command
llama-server -m gary-qwen36-27b-v4-Q4_K_M.gguf -ngl 99 -c 73728 -np 1 --cache-type-k bf16 --cache-type-v bf16 --jinja --flash-attn on --batch-size 2048 --ubatch-size 512 --threads 8 --threads-batch 16 --temp 0.1 --reasoning-format deepseek --reasoning-budget 1024 --no-context-shift --ctx-checkpoints 0 --checkpoint-every-n-tokens -1
localmaxxing · run id
cmoerh82s0001l404jm347jzo
localmaxxing · tokenized · arguments
llama-server, -m, gary-qwen36-27b-v4-Q4_K_M.gguf, -ngl, 99, -c, 73728, -np, 1, --cache-type-k, bf16, --cache-type-v, bf16, --jinja, --flash-attn, on, --batch-size, 2048, --ubatch-size, 512, --threads, 8, --threads-batch, 16, --temp, 0.1, --reasoning-format, deepseek, --reasoning-budget, 1024, --no-context-shift, --ctx-checkpoints, 0, --checkpoint-every-n-tokens, -1
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/cmoerh82s0001l404jm347jzo