Local AI Registry

Recipe

muse-glimmer-30b-rocmfp4-fast-radeon-ai-pro-r9700-32gb-llama-cpp-tp3

muse-glimmer-30b-rocmfp4-fast-radeon-ai-pro-r9700-32gb-llama-cpp-tp3
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
b10344-11-gdd1ea5243
Accelerators
3
Tensor parallel
3
Context tokens
1,923
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/meta-models/Muse-Glimmer-30B
Repository
meta-models/Muse-Glimmer-30B
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/cmso8xev00005mp01f3uco1go

Observed source tokens

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

  1. /home/mikekey/.config/llama-swap/gpu-order.sh
  2. exec
  3. /home/mikekey/Experiments/ROCmFPX/build-rdna4/bin/llama-server
  4. --port
  5. ${PORT}
  6. --host
  7. 0.0.0.0
  8. --no-webui
  9. -ngl
  10. 99
  11. --jinja
  12. --flash-attn
  13. on
  14. --cache-reuse
  15. 256
  16. --prio
  17. 3
  18. -m
  19. /home/mikekey/models/gguf/meta/Muse-Glimmer-30B-ROCMFP4_FAST.gguf
  20. -md
  21. /home/mikekey/models/gguf/meta/Muse-Glimmer-30B-dflash-draft-Q8_0.gguf
  22. --spec-type
  23. draft-dflash
  24. --spec-draft-p-min
  25. 0.3
  26. --spec-draft-n-max
  27. 16
  28. --chat-template-kwargs
  29. {"reasoning_strength":"zero"}
  30. --reasoning-format
  31. auto
  32. --temp
  33. 0.6
  34. --top-p
  35. 0.95
  36. --top-k
  37. 20
  38. --min-p
  39. 0.0
  40. --repeat-penalty
  41. 1.0
  42. -ctk
  43. f16
  44. -ctv
  45. f16
  46. -ts
  47. 1/0/0
  48. -np
  49. 1
  50. -c
  51. 131072
FlagValue
--port${PORT}
--host0.0.0.0
-ngl99
--flash-attnon
--cache-reuse256
--prio3
-m/home/mikekey/models/gguf/meta/Muse-Glimmer-30B-ROCMFP4_FAST.gguf
-md/home/mikekey/models/gguf/meta/Muse-Glimmer-30B-dflash-draft-Q8_0.gguf
--spec-typedraft-dflash
--spec-draft-p-min0.3
--spec-draft-n-max16
--chat-template-kwargs{"reasoning_strength":"zero"}
--reasoning-formatauto
--temp0.6
--top-p0.95
--top-k20
--min-p0.0
--repeat-penalty1.0
-ctkf16
-ctvf16
-ts1/0/0
-np1
-c131072

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
11,9231,116.862.61,263.4observedmuse-glimmer-30b-rocmfp4-fast-radeon-ai-pro-r9700-32gb-llama-cpp-tp3-sweep

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

hardware count
3
hardware id
radeon-ai-pro-r9700-32gb
id
muse-glimmer-30b-rocmfp4-fast-radeon-ai-pro-r9700-32gb-llama-cpp-tp3
model instance id
meta-models-muse-glimmer-30b--rocmfp4-fast
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
muse-glimmer-30b-rocmfp4-fast-radeon-ai-pro-r9700-32gb-llama-cpp-tp3-sweep
status
candidate

capabilities

engine

name
llama.cpp
version
b10344-11-gdd1ea5243

serving

max concurrency
1
max context tokens
1,923
tensor parallel
3
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
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
rocm
localmaxxing · batch size
1
localmaxxing · hardware label
Radeon AI Pro R9700
localmaxxing · notes
llama.cpp (ROCmFPX fork) ROCm, single R9700, FP4_FAST 13.8GiB + DFlash draft Q8_0 2.6GiB. Greedy, ~1.4k-tok code prompt. spec: draft-dflash p-min 0.3 n-max 16, reasoning_strength=zero. Median of 3.
localmaxxing · run id
cmso8xev00005mp01f3uco1go
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/cmso8xev00005mp01f3uco1go ↗