Recipe

qwen3-6-35b-a3b-ud-q3-k-xl-rtx-3090-24gb-llama-cpp-tp1

qwen3-6-35b-a3b-ud-q3-k-xl-rtx-3090-24gb-llama-cpp-tp1

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

Record

Status
candidate
Source
localmaxxing
Engine
llama.cpp
Engine version
b9102 (c8f8e2364, gg/spec-mtp-experiments)
Accelerators
1
Tensor parallel
1
Context tokens
163,840
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/unsloth/Qwen3.6-35B-A3B-MTP-GGUF
Repository
unsloth/Qwen3.6-35B-A3B-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/cmp4dpnef00c4pc01awn43jqk

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. --flash-attn
  3. on
  4. -m
  5. Qwen3.6-35B-A3B-UD-Q3_K_XL.gguf
  6. -c
  7. 163840
  8. --port
  9. 8080
  10. --parallel
  11. 1
  12. -ngl
  13. 99
  14. --mlock
  15. --host
  16. 0.0.0.0
  17. --no-mmap
  18. --cache-type-k
  19. q8_0
  20. --cache-type-v
  21. q8_0
  22. --cache-ram
  23. 0
  24. --spec-type
  25. mtp
  26. --spec-draft-type-k
  27. q8_0
  28. --spec-draft-type-v
  29. q8_0
FlagValue
--flash-attnon
-mQwen3.6-35B-A3B-UD-Q3_K_XL.gguf
-c163840
--port8080
--parallel1
-ngl99
--host0.0.0.0
--cache-type-kq8_0
--cache-type-vq8_0
--cache-ram0
--spec-typemtp
--spec-draft-type-kq8_0
--spec-draft-type-vq8_0

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
1163,840512.5195.124observedqwen3-6-35b-a3b-ud-q3-k-xl-rtx-3090-24gb-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-3090-24gb
id
qwen3-6-35b-a3b-ud-q3-k-xl-rtx-3090-24gb-llama-cpp-tp1
model instance id
unsloth-qwen3-6-35b-a3b-mtp-gguf--ud-q3-k-xl
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-35b-a3b-ud-q3-k-xl-rtx-3090-24gb-llama-cpp-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, unsloth/Qwen3.6-35B-A3B-MTP-GGUF, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 163840
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:b2497f8834f5ecb4e38530f6bf2734b8e0be107ff48e4720145911c86930f2ce
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
b9102 (c8f8e2364, gg/spec-mtp-experiments)

serving

max concurrency
1
max context tokens
163,840
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
draft launch.entrypoint · provenance · captured at
2026-09-01T01:41:26Z
draft launch.entrypoint · reason linux-amd64-container-config-entrypoint
draft launch.entrypoint · state known
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
cuda
localmaxxing · batch size
1
localmaxxing · hardware label
RTX 3090
localmaxxing · notes
llama.cpp MTP build from gg/spec-mtp-experiments branch. MTP active: 40/40 draft acceptance. Q8_0 KV cache. --cache-ram 0 required for large context. Full GPU offload (-ngl 99).
localmaxxing · observed command
llama-server --flash-attn on -m Qwen3.6-35B-A3B-UD-Q3_K_XL.gguf -c 163840 --port 8080 --parallel 1 -ngl 99 --mlock --host 0.0.0.0 --no-mmap --cache-type-k q8_0 --cache-type-v q8_0 --cache-ram 0 --spec-type mtp --spec-draft-type-k q8_0 --spec-draft-type-v q8_0
localmaxxing · run id
cmp4dpnef00c4pc01awn43jqk
localmaxxing · tokenized · arguments
llama-server, --flash-attn, on, -m, Qwen3.6-35B-A3B-UD-Q3_K_XL.gguf, -c, 163840, --port, 8080, --parallel, 1, -ngl, 99, --mlock, --host, 0.0.0.0, --no-mmap, --cache-type-k, q8_0, --cache-type-v, q8_0, --cache-ram, 0, --spec-type, mtp, --spec-draft-type-k, q8_0, --spec-draft-type-v, q8_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/cmp4dpnef00c4pc01awn43jqk