Recipe

deepseek-v4-flash-q2-k-ryzen-ai-max-plus-395-128gb-dwarfstar-tp1

deepseek-v4-flash-q2-k-ryzen-ai-max-plus-395-128gb-dwarfstar-tp1

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

Record

Status
candidate
Source
localmaxxing
Engine
dwarfstar
Accelerators
1
Tensor parallel
1
Context tokens
16,384
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash
Repository
deepseek-ai/DeepSeek-V4-Flash
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

dwarfstar

Evidence only · candidate · reference

Candidate evidence — not a Run contract

Source
https://www.localmaxxing.com/en/runs/cmrgn4jtk005hmj01022ts8t3

Observed source tokens

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

  1. ds4-server
  2. -m
  3. DeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed.gguf
  4. --mtp
  5. DeepSeek-V4-Flash-MTP-Q4K-Q8_0-F32.gguf
  6. --ctx
  7. 16384
  8. --host
  9. 0.0.0.0
  10. --port
  11. 8000
FlagValue
-mDeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed.gguf
--mtpDeepSeek-V4-Flash-MTP-Q4K-Q8_0-F32.gguf
--ctx16384
--host0.0.0.0
--port8000

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
116,3849015.6observeddeepseek-v4-flash-q2-k-ryzen-ai-max-plus-395-128gb-dwarfstar-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
ryzen-ai-max-plus-395-128gb
id
deepseek-v4-flash-q2-k-ryzen-ai-max-plus-395-128gb-dwarfstar-tp1
model instance id
deepseek-ai-deepseek-v4-flash--q2-k
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
deepseek-v4-flash-q2-k-ryzen-ai-max-plus-395-128gb-dwarfstar-tp1-sweep
status
candidate

capabilities

engine

name
dwarfstar

serving

max concurrency
1
max context tokens
16,384
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
rocm
localmaxxing · hardware label
Ryzen AI Max 395
localmaxxing · notes
DS4 (antirez DwarfStar) via kyuz0/strix-halo-ds4-toolbox:rocm-7.2.4, rootless podman on Framework Desktop (Strix Halo gfx1151, 120G GTT). antirez/deepseek-v4-gguf Q2/Q4 hybrid quant (97G: layers 37-42 experts Q4_K, other expert layers IQ2_XXS gate/up + Q2_K down, attn/shared-expert/output Q8) + MTP draft for spec decode. llama.cpp's V4 path hangs on prefill on this hw; DS4 is the working engine for DeepSeek-V4 GGUFs here.
localmaxxing · observed command
ds4-server -m DeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed.gguf --mtp DeepSeek-V4-Flash-MTP-Q4K-Q8_0-F32.gguf --ctx 16384 --host 0.0.0.0 --port 8000
localmaxxing · run id
cmrgn4jtk005hmj01022ts8t3
localmaxxing · tokenized · arguments
ds4-server, -m, DeepSeek-V4-Flash-Layers37-42Q4KExperts-OtherExpertLayersIQ2XXSGateUp-Q2KDown-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix-fixed.gguf, --mtp, DeepSeek-V4-Flash-MTP-Q4K-Q8_0-F32.gguf, --ctx, 16384, --host, 0.0.0.0, --port, 8000
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/cmrgn4jtk005hmj01022ts8t3