Recipe
nvidia-nemotron-3-5-lightning-30b-a3b-bf16-unsloth-dynamic-q4-k-m-intel-arc-pro-b70-32gb-llama-cpp-tp1
nvidia-nemotron-3-5-lightning-30b-a3b-bf16-unsloth-dynamic-q4-k-m-intel-arc-pro-b70-32gb-llama-cpp-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Engine version
- 9fee29e9435f; oneAPI 2026.1.1
- 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/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF- Repository
- unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF
- 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
Observed source tokens
Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.
llama-server-mNVIDIA-Nemotron-3.5-Lightning-30B-A3B-UD-Q4_K_M.gguf--reasoningoff-c8192-ctkf16-ctvf16--deviceSYCL0-ngl99-faauto--parallel1--cache-ram0--ctx-checkpoints0--fitoff
| Flag | Value |
|---|---|
-m | NVIDIA-Nemotron-3.5-Lightning-30B-A3B-UD-Q4_K_M.gguf |
--reasoning | off |
-c | 8192 |
-ctk | f16 |
-ctv | f16 |
--device | SYCL0 |
-ngl | 99 |
-fa | auto |
--parallel | 1 |
--cache-ram | 0 |
--ctx-checkpoints | 0 |
--fit | off |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 8,192 | — | 72.2 | — | observed | nvidia-nemotron-3-5-lightning-30b-a3b-bf16-unsloth-dynamic-q4-k-m-intel-arc-pro-b70-32gb-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
- intel-arc-pro-b70-32gb
- id
- nvidia-nemotron-3-5-lightning-30b-a3b-bf16-unsloth-dynamic-q4-k-m-intel-arc-pro-b70-32gb-llama-cpp-tp1
- model instance id
- unsloth-nvidia-nemotron-3-5-lightning-30b-a3b-gguf--unsloth-dynamic-q4-k-m
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- nvidia-nemotron-3-5-lightning-30b-a3b-bf16-unsloth-dynamic-q4-k-m-intel-arc-pro-b70-32gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- 9fee29e9435f; oneAPI 2026.1.1
serving
- max concurrency
- 1
- max context tokens
- 8,192
- 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
- xpu
- localmaxxing · hardware label
- Intel Arc Pro B70
- localmaxxing · notes
- Nemotron 3.5 Lightning 30B-A3B UD-Q4_K_M target-only on one Intel Arc Pro B70 with stock pinned llama.cpp SYCL. Fresh cold 12-prompt suite, 512-token responses, conventional 99-interval median, cache_prompt=false and cached_tokens=0; second fresh server measured 72.035976 tok/s. Model pin, commands, and measured decode/prefill through 32K: https://neural.download/models/nemotron-35-lightning-30b-a3b-b70.html and https://github.com/steveseguin/b70-optimization-lab/tree/main/repro/nemotron-35-lightning-30b-a3b-b70. MoE uses about 3B active parameters per token.
- localmaxxing · observed command
- llama-server -m NVIDIA-Nemotron-3.5-Lightning-30B-A3B-UD-Q4_K_M.gguf --reasoning off -c 8192 -ctk f16 -ctv f16 --device SYCL0 -ngl 99 -fa auto --parallel 1 --cache-ram 0 --ctx-checkpoints 0 --fit off
- localmaxxing · run id
- cmt9m8hgp00ysli017rhfj46x
- localmaxxing · tokenized · arguments
- llama-server, -m, NVIDIA-Nemotron-3.5-Lightning-30B-A3B-UD-Q4_K_M.gguf, --reasoning, off, -c, 8192, -ctk, f16, -ctv, f16, --device, SYCL0, -ngl, 99, -fa, auto, --parallel, 1, --cache-ram, 0, --ctx-checkpoints, 0, --fit, off
- localmaxxing · tokenized · fidelity
- faithful
provenance
- captured at
- 2026-08-30T09:10:02Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-30T09:10:02Z | normalized-recipe | www.localmaxxing.com/en/runs/cmt9m8hgp00ysli017rhfj46x ↗ |