Recipe
deepseek-v4-flash-0731-iq2-xxs-radeon-ai-pro-r9700-32gb-lucebox-tp1-7ibvc9zx
deepseek-v4-flash-0731-iq2-xxs-radeon-ai-pro-r9700-32gb-lucebox-tp1-7ibvc9zxObserved LocalMaxxing result. It is evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- lucebox
- Engine version
- c994209
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 8,192
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731- Repository
- deepseek-ai/DeepSeek-V4-Flash-0731
- 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
lucebox
Evidence only · candidate · reference
Candidate evidence — not a Run contract
No tokenized launch fields. This is measured or documented compatibility, not a Docker launch.
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 8,192 | 196.1 | 59.4 | 2,003.9 | observed | deepseek-v4-flash-0731-iq2-xxs-radeon-ai-pro-r9700-32gb-lucebox-tp1-7ibvc9zx-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
- radeon-ai-pro-r9700-32gb
- id
- deepseek-v4-flash-0731-iq2-xxs-radeon-ai-pro-r9700-32gb-lucebox-tp1-7ibvc9zx
- model instance id
- deepseek-ai-deepseek-v4-flash-0731--iq2-xxs
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- deepseek-v4-flash-0731-iq2-xxs-radeon-ai-pro-r9700-32gb-lucebox-tp1-7ibvc9zx-sweep
- status
- candidate
capabilities
engine
- name
- lucebox
- version
- c994209
serving
- max concurrency
- 1
- max context tokens
- 8,192
- tensor parallel
- 1
Provenance & metadata (3)
facts
- capabilities.chat · provenance · captured at
- 2026-08-27T06:04:13.773Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
- capabilities.reasoning · provenance · captured at
- 2026-08-27T06:04:13.773Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
- capabilities.tools · provenance · captured at
- 2026-08-27T06:04:13.773Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
- capabilities.vision · provenance · captured at
- 2026-08-27T06:04:13.773Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
- engine.graph mode · provenance · captured at
- 2026-08-27T06:04:13.773Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
- serving.kv cache tokens · provenance · captured at
- 2026-08-27T06:04:13.773Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
metadata
- localmaxxing · hardware label
- Radeon AI Pro R9700
- localmaxxing · notes
- Corrected PR #640 qualification: exact c994209 dual-architecture gfx1151+gfx1201 build; DeepSeek V4 Flash ROCmFP2 target plus DSpark denseF16 draft, q=4/top-k4; R9700 hip:0 plus Strix Halo 8060S hip:1 over Oculink peer access; calibrated hotness profile; actual 10200 MiB expert budget; grouped heterogeneous MMID prefill enabled with expert-major threshold 64; PR #647 cold-input-first disabled; both GGML_BATCH_PEER_COPIES and GGML_CUDA_BATCH_PEER_COPIES disabled; sparse prefill, chunk 2048, context 8192. Protocol: one discarded warmup plus three measured streaming requests at temperature 0, 393 prompt tokens and 128 output tokens. All measured outputs shared SHA-256 0f785a7ffa406498aafb14553966eaed0f52220fed0f7cc016b66921d104d194. Activation trace proved device-resident heterogeneous expert-major prefill. TTFT and rates are client monotonic measurements. Peak VRAM is R9700 workload allocation: max /sys/class/drm/card1/device/mem_info_vram_used minus stable pre-launch driver baseline (27076444160 - 59936768 bytes). Structured hardware names the discrete R9700; Halo participation is disclosed here. Short-context benchmark profile, not a general production configuration.
- localmaxxing · revision unpinned
- Yes
- localmaxxing · run id
- cmta1du8p01anli017ibvc9zx
provenance
- captured at
- 2026-08-27T06:04:13.773Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-27T06:04:13.773Z | normalized-recipe | www.localmaxxing.com/en/runs/cmta1du8p01anli017ibvc9zx ↗ |