Recipe
glm52-nvfp4-reap-nu176-rtxpro6000-vllm-tp4
glm52-nvfp4-reap-nu176-rtxpro6000-vllm-tp4Pinned-source non-uniform REAP NU176 candidate for four RTX PRO 6000 Blackwell GPUs. The source reports 262K context, concurrency two, MTP4, and full/piecewise CUDA graphs, but model and image digest pins plus current index acceptance are still missing.
Record
- Status
- candidate
- Source
- 0xsero
- Engine
- vllm
- Engine version
- b12x nonuniform fork 20260705; exact commit unreported
- Graph
- full-and-piecewise
- Accelerators
- 4
- Tensor parallel
- 4
- Context tokens
- 262,144
- Max concurrency
- 2
- chat
- yes
- reasoning
- yes
- tools
- yes
- vision
- no
Hugging Face model card
Identity
https://huggingface.co/0xSero/GLM-5.2-REAP-NU176-526B- Repository
- 0xSero/GLM-5.2-REAP-NU176-526B
- 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.
Compose configuration
vllm
Container · candidate · docker compose
Candidate evidence — not a Run contract
- Compose file
compose.yml- Port
- 8000
Environment
| Variable | Value |
|---|---|
CUDA_VISIBLE_DEVICES | 0,1,2,3 |
NCCL_P2P_DISABLE | 1 |
VLLM_USE_B12X_MOE | 1 |
VLLM_USE_B12X_SPARSE_INDEXER | 1 |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | — | — | 77 | — | historical | glm52-nvfp4-reap-nu176-rtxpro6000-vllm-tp4-sweep |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- hardware count
- 4
- hardware id
- rtx-pro-6000-blackwell-96gb
- id
- glm52-nvfp4-reap-nu176-rtxpro6000-vllm-tp4
- model instance id
- 0xsero-glm-5-2-reap-nu176-526b--nvfp4-non-uniform-reap
- recipe source
- 0xsero
- schema version
- local-ai-registry/v1
- speed sweep ids
- glm52-nvfp4-reap-nu176-rtxpro6000-vllm-tp4-sweep
- status
- candidate
capabilities
- chat
- Yes
- reasoning
- Yes
- tools
- Yes
- vision
- No
engine
- graph mode
- full-and-piecewise
- name
- vllm
- version
- b12x nonuniform fork 20260705; exact commit unreported
serving
- max concurrency
- 2
- max context tokens
- 262,144
- tensor parallel
- 4
Provenance & metadata (3)
facts
- 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
provenance
- captured at
- 2026-08-27T06:04:13.773Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-27T06:04:13.773Z | normalized-recipe | github.com/0xSero/local-ai-registry ↗ |