Recipe

mimo-v2-5-q4-k-m-gguf-dgx-spark-gb10-128gb-llama-cpp-tp1

mimo-v2-5-q4-k-m-gguf-dgx-spark-gb10-128gb-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
1e5ad35-2026-05-10-built-from-master
Accelerators
1
Tensor parallel
1
Context tokens
1,000,000
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/bartowski/MiMo-V2.5-GGUF
Repository
bartowski/MiMo-V2.5-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/cmozdo5940030q201948atk3q

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. MiMo-V2.5-Q4_K_M-00001-of-00005.gguf
  4. --host
  5. 0.0.0.0
  6. --port
  7. 10000
  8. --ctx-size
  9. 1000000
  10. -np
  11. 4
  12. --n-gpu-layers
  13. 999
  14. --rpc
  15. 192.168.101.1:50052
  16. --threads
  17. 8
  18. --threads-batch
  19. 8
  20. -fa
  21. 1
  22. --batch-size
  23. 2048
  24. --ubatch-size
  25. 512
FlagValue
-mMiMo-V2.5-Q4_K_M-00001-of-00005.gguf
--host0.0.0.0
--port10000
--ctx-size1000000
-np4
--n-gpu-layers999
--rpc192.168.101.1:50052
--threads8
--threads-batch8
-fa1
--batch-size2048
--ubatch-size512

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
11,000,00016.1606observedmimo-v2-5-q4-k-m-gguf-dgx-spark-gb10-128gb-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
dgx-spark-gb10-128gb
id
mimo-v2-5-q4-k-m-gguf-dgx-spark-gb10-128gb-llama-cpp-tp1
model instance id
bartowski-mimo-v2-5-gguf--q4-k-m-gguf
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
mimo-v2-5-q4-k-m-gguf-dgx-spark-gb10-128gb-llama-cpp-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
-hf, bartowski/MiMo-V2.5-GGUF, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 1000000
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
1e5ad35-2026-05-10-built-from-master

serving

max concurrency
1
max context tokens
1,000,000
tensor parallel
1
Provenance & metadata (3)

facts

capabilities.chat · provenance · captured at
2026-08-30T09:26:09Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
capabilities.reasoning · provenance · captured at
2026-08-30T09:26:09Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
capabilities.tools · provenance · captured at
2026-08-30T09:26:09Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
capabilities.vision · provenance · captured at
2026-08-30T09:26:09Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
engine.graph mode · provenance · captured at
2026-08-30T09:26:09Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
serving.kv cache tokens · provenance · captured at
2026-08-30T09:26:09Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown

metadata

localmaxxing · hardware label
GB10 Grace Blackwell
localmaxxing · notes
MiMo V2.5 (309.77B params, 51 layers, MoE) Q4_K_M GGUF on dual NVIDIA DGX Spark cluster via llama.cpp RPC backend. tokSOut=16.08 t/s median c=1 (5 runs max_tokens=256, mean 16.4, stdev 0.91). Aggregate at c=4 concurrent ~32 t/s (8.15 per stream). TTFR 606ms median. Context 1M (5x what M2.7 supports on same cluster). Built from ggml-org/llama.cpp master commit 1e5ad35 with GGML_RPC=ON GGML_CUDA=ON sm_121a target. Includes PR 22493 (model arch May 7) + PR 22812 (FA kernels May 9). Layer split: 50/52 layers offloaded, CUDA0 80.6 GiB on Spark 2 + RPC0 98.3 GiB on Spark 1 over QSFP56 200G. AesSedai vision branch not used (text-only).
localmaxxing · observed command
llama-server -m MiMo-V2.5-Q4_K_M-00001-of-00005.gguf --host 0.0.0.0 --port 10000 --ctx-size 1000000 -np 4 --n-gpu-layers 999 --rpc 192.168.101.1:50052 --threads 8 --threads-batch 8 -fa 1 --batch-size 2048 --ubatch-size 512
localmaxxing · run id
cmozdo5940030q201948atk3q
localmaxxing · tokenized · arguments
llama-server, -m, MiMo-V2.5-Q4_K_M-00001-of-00005.gguf, --host, 0.0.0.0, --port, 10000, --ctx-size, 1000000, -np, 4, --n-gpu-layers, 999, --rpc, 192.168.101.1:50052, --threads, 8, --threads-batch, 8, -fa, 1, --batch-size, 2048, --ubatch-size, 512
localmaxxing · tokenized · fidelity
faithful

provenance

captured at
2026-08-30T09:26:09Z

sources

captured atkindurl
2026-08-30T09:26:09Znormalized-recipewww.localmaxxing.com/en/runs/cmozdo5940030q201948atk3q