UAPI Grok Hardware Autorouter Cloud GPU Runbook
Grok-owned hardware inventory, autorouter, cloud GPU, 140T/NPU research, and friendly route tasks.
Promotion Metadata
- Source feedback id:
80 - Source feedback ids:
80 - Promotion request id:
15 - Feedback category:
consider - Feedback source:
ai-lane - Feedback created at:
2026-06-22T16:31:04.363192+00:00 - Target slug:
[REDACTED:high-entropy] - Review state:
review - Reviewer note: AI lane auto-promoted redacted feedback into internal catalog context.
Source Feedback Body
Feedback 80
- Product:
uapi - Category:
consider - Related slug:
[REDACTED:high-entropy] - Source:
ai-lane - Created at:
2026-06-22T16:31:04.363192+00:00
title: UAPI Grok Hardware Autorouter Cloud GPU Runbook project: uapi kind: runbook state: review tags: [docs, runbook, uapi, grok, hardware, autorouter, gpu, npu] source_plan_title: UAPI Integrated Operating Plan v2
UAPI Grok Hardware Autorouter Cloud GPU Runbook
Owner: Grok. Mission: produce the verified hardware and routing evidence that UAPI needs before it can place jobs on lab, NUC, 140T, NPU, SN850X, Colab, or cloud GPU lanes.
Access contract: Endpoint: http://10.0.0.121:8080. Token: set UAI_EXPLORER_AI_API_KEY out of band. Header uses Bearer token from that variable. Never print secrets. Fetch master program control plus the main integrated UAPI plan. Lab devices are not local.
Required hardware review: - Confirm exact NUC hardware: CPU SKU, Intel Arc 140T or 140-series details, onboard NPU generation, OS, drivers, OpenVINO/IPEX/llama.cpp support, shared memory/VRAM behavior, thermals, power, and stability. - Verify whether the NUC CPU must remain reserved for Conductor/Future Foundry. - Re-inventory SN850X. If it is no longer tied to Gen4/VPS-Cortex duty, model it as UAPI-served storage/capacity only after facts are verified. - Verify VPS/Cortex host reality, lab-hosted GPU routes, and Colab/cloud GPU burst routes.
140T/NPU research pass: treat 140T plus NPU as a research target for faster local Gemma/SLM lanes, not production until benchmarked. Compare GPU, NPU, CPU, and hybrid placement for small Gemma/SLM chat, embeddings, rerankers, intent classifiers, voice/PA helper models, VLMs, batch inference, and background low-power inference.
Adjacent product review seeds by project name: OpenVINO GenAI, OpenVINO Notebooks Gemma 4 notebook, OpenVINO GenAI NPU docs, Intel AI Playground, IPEX-LLM, OpenArc, NoLlama, llama.cpp Intel Arc/NPU/OpenVINO discussion, and Intel NPU Acceleration Library as archived historical reference only. Use web and GitHub search to resolve current URLs. Do not fork or import code until license, maintenance, security, and fit are reviewed.
Deliverables: hardware inventory table, benchmark plan, route capability matrix, friendly route names such as local.nuc.gemma.npu, local.nuc.gemma.140t, lab.cortex.memory, cloud.colab.gpu.burst, and PlacementDecision evidence rules.
Benchmark metrics: first-token latency, tokens/sec, model load time, memory use, power, thermals, stability, driver pain, API compatibility, fallback behavior.
Grok prompt starter: You are Grok for UAPI hardware and autorouter. Use Explorer first, web/GitHub for current hardware/backend research, and publish verified findings with citations, audit ids, assumptions, and no secrets.