Local AI Registry

Recipe

qwen3-8-27b-ud-q2-k-xl-rtx-5070-12gb-llama-cpp-tp1

qwen3-8-27b-ud-q2-k-xl-rtx-5070-12gb-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
b10449
Accelerators
1
Tensor parallel
1
Context tokens
8,192
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/unsloth/Qwen3.8-27B-GGUF
Repository
unsloth/Qwen3.8-27B-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/cmt3jp2cn0njpmv01t5u8m6wd

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. --model
  3. Qwen3.8-27B-UD-Q2_K_XL.gguf
  4. --host
  5. 127.0.0.1
  6. --port
  7. 8080
  8. --no-ui
  9. --ctx-size
  10. 8192
  11. --batch-size
  12. 256
  13. --ubatch-size
  14. 64
  15. --flash-attn
  16. on
  17. --cache-type-k
  18. q4_0
  19. --cache-type-v
  20. q4_0
  21. --spec-draft-type-k
  22. q4_0
  23. --spec-draft-type-v
  24. q4_0
  25. --spec-type
  26. draft-mtp
  27. --spec-draft-n-max
  28. 1
  29. --device
  30. CUDA0
  31. --n-gpu-layers
  32. all
  33. --kv-offload
  34. --parallel
  35. 1
  36. --fit
  37. off
  38. --jinja
  39. --chat-template-file
  40. fixed-v22.3-chat_template.jinja
  41. --reasoning
  42. off
  43. --no-reasoning-preserve
  44. --reasoning-format
  45. deepseek
  46. --predict
  47. 512
  48. --offline
FlagValue
--modelQwen3.8-27B-UD-Q2_K_XL.gguf
--host127.0.0.1
--port8080
--ctx-size8192
--batch-size256
--ubatch-size64
--flash-attnon
--cache-type-kq4_0
--cache-type-vq4_0
--spec-draft-type-kq4_0
--spec-draft-type-vq4_0
--spec-typedraft-mtp
--spec-draft-n-max1
--deviceCUDA0
--n-gpu-layersall
--parallel1
--fitoff
--chat-template-filefixed-v22.3-chat_template.jinja
--reasoningoff
--reasoning-formatdeepseek
--predict512

Environment

VariableValue
CUDA_VISIBLE_DEVICES0

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
18,192727.370.8704observedqwen3-8-27b-ud-q2-k-xl-rtx-5070-12gb-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
rtx-5070-12gb
id
qwen3-8-27b-ud-q2-k-xl-rtx-5070-12gb-llama-cpp-tp1
model instance id
unsloth-qwen3-8-27b-gguf--ud-q2-k-xl
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-8-27b-ud-q2-k-xl-rtx-5070-12gb-llama-cpp-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, unsloth/Qwen3.8-27B-GGUF, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 8192
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
b10449

serving

max concurrency
1
max context tokens
8,192
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 · backend
cuda
localmaxxing · hardware label
RTX 5070
localmaxxing · notes
WumboJetsII RTX 5070 12GB. Unsloth Qwen3.8-27B UD-Q2_K_XL, pinned revision, froggeric FIXED v22.3 external template, llama.cpp b10449 CUDA. Full GPU model/KV/recurrent state, parallel=1, fit off. Fresh LocalMaxxing-compatible run. Native MTP nmax=1. llama-server /completion ignore_eos 512/512; 1 warmup + 3 reps; submitted mean not max (69.28-71.62 tok/s). Draft accept 254/257, 255/256, 246/264.
localmaxxing · observed command
CUDA_VISIBLE_DEVICES=0 llama-server --model Qwen3.8-27B-UD-Q2_K_XL.gguf --host 127.0.0.1 --port 8080 --no-ui --ctx-size 8192 --batch-size 256 --ubatch-size 64 --flash-attn on --cache-type-k q4_0 --cache-type-v q4_0 --spec-draft-type-k q4_0 --spec-draft-type-v q4_0 --spec-type draft-mtp --spec-draft-n-max 1 --device CUDA0 --n-gpu-layers all --kv-offload --parallel 1 --fit off --jinja --chat-template-file fixed-v22.3-chat_template.jinja --reasoning off --no-reasoning-preserve --reasoning-format deepseek --predict 512 --offline
localmaxxing · run id
cmt3jp2cn0njpmv01t5u8m6wd
localmaxxing · tokenized · arguments
llama-server, --model, Qwen3.8-27B-UD-Q2_K_XL.gguf, --host, 127.0.0.1, --port, 8080, --no-ui, --ctx-size, 8192, --batch-size, 256, --ubatch-size, 64, --flash-attn, on, --cache-type-k, q4_0, --cache-type-v, q4_0, --spec-draft-type-k, q4_0, --spec-draft-type-v, q4_0, --spec-type, draft-mtp, --spec-draft-n-max, 1, --device, CUDA0, --n-gpu-layers, all, --kv-offload, --parallel, 1, --fit, off, --jinja, --chat-template-file, fixed-v22.3-chat_template.jinja, --reasoning, off, --no-reasoning-preserve, --reasoning-format, deepseek, --predict, 512, --offline
localmaxxing · tokenized · environment · CUDA VISIBLE DEVICES
0
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/cmt3jp2cn0njpmv01t5u8m6wd ↗