Recipe

deepseek-v4-flash-iq2-xxs-asym-moe-apple-m5-max-128gb-llama-cpp-tp1

deepseek-v4-flash-iq2-xxs-asym-moe-apple-m5-max-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
ds4 80ebbc3
Accelerators
1
Tensor parallel
1
Context tokens
32,768
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

llama.cpp

Evidence only · candidate · reference

Candidate evidence — not a Run contract

Source
https://www.localmaxxing.com/en/runs/cmqpiopug00ouqo01gtohwf4l

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. --model
  3. ds4flash.gguf
  4. --ctx
  5. 32768
  6. (antirez/ds4
  7. DwarfStar
  8. native
  9. Metal
  10. engine;
  11. q2-imatrix
  12. IQ2_XXS
  13. GGUF
  14. from
  15. antirez/deepseek-v4-gguf;
  16. in-memory
  17. KV,
  18. no
  19. MTP)
FlagValue
--modelds4flash.gguf
--ctx32768

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
132,76832026.3observeddeepseek-v4-flash-iq2-xxs-asym-moe-apple-m5-max-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
apple-m5-max-128gb
id
deepseek-v4-flash-iq2-xxs-asym-moe-apple-m5-max-128gb-llama-cpp-tp1
model instance id
deepseek-ai-deepseek-v4-flash--iq2-xxs-asym-moe
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
deepseek-v4-flash-iq2-xxs-asym-moe-apple-m5-max-128gb-llama-cpp-tp1-sweep
status
candidate

capabilities

engine

name
llama.cpp
version
ds4 80ebbc3

serving

max concurrency
1
max context tokens
32,768
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 · hardware label
M5 Max
localmaxxing · notes
antirez/ds4 (DwarfStar) custom Metal engine for DeepSeek-V4-Flash (filed as llama.cpp per cell convention; not actual llama.cpp). IQ2_XXS asymmetric MoE quant. Decode measured at the 32k context frontier via ds4-bench, greedy, in-memory KV (no disk KV cache, no MTP): gen 26.33 t/s, prefill 319.95 t/s. Apple M5 Max 128GB macOS 27.0.
localmaxxing · observed command
ds4-server --model ds4flash.gguf --ctx 32768 (antirez/ds4 "DwarfStar" native Metal engine; q2-imatrix IQ2_XXS GGUF from antirez/deepseek-v4-gguf; in-memory KV, no MTP)
localmaxxing · run id
cmqpiopug00ouqo01gtohwf4l
localmaxxing · tokenized · arguments
ds4-server, --model, ds4flash.gguf, --ctx, 32768, (antirez/ds4, DwarfStar, native, Metal, engine;, q2-imatrix, IQ2_XXS, GGUF, from, antirez/deepseek-v4-gguf;, in-memory, KV,, no, MTP)
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/cmqpiopug00ouqo01gtohwf4l