Recipe
carnice-9b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1
carnice-9b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 512
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/kai-os/Carnice-9b-GGUF- Repository
- kai-os/Carnice-9b-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-bench-m-p512-n128-ngl32-smlayer-faon-ctkf16-ctvf16-r5
| Flag | Value |
|---|---|
-p | 512 |
-n | 128 |
-ngl | 32 |
-sm | layer |
-fa | on |
-ctk | f16 |
-ctv | f16 |
-r | 5 |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 512 | 1,671.5 | 56.4 | 483.1 | observed | carnice-9b-q4-k-m-rtx-3060-12gb-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
- rtx-3060-12gb
- id
- carnice-9b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1
- model instance id
- kai-os-carnice-9b-gguf--q4-k-m
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- carnice-9b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
draft launch
- accelerator backend
- nvidia
- arguments
- -hf, kai-os/Carnice-9b-GGUF:Q4_K_M, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 512
- container port
- 8,080
- environment · LLAMA CACHE
- /root/.cache/huggingface
- host port
- 8,080
- image
- ghcr.io/ggml-org/llama.cpp:server-cuda12-b10481@sha256:b2497f8834f5ecb4e38530f6bf2734b8e0be107f0f0857e259672d1cb85b71c2
- ipc
- host
- kind
- docker
- shm size
- 16g
- synthesized · generated at
- 2026-08-31T22:12:17Z
- synthesized · image provenance
- gemma-4-12b-q4-k-m-rtx-3060-12gb-llama-cpp-tp1
- synthesized · template
- llama-cpp-server-v1
mounts
| read only | target |
|---|---|
| No | /root/.cache/huggingface |
engine
- name
- llama.cpp
serving
- max concurrency
- 1
- max context tokens
- 512
- 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
- cuda
- localmaxxing · hardware label
- RTX 3060
- localmaxxing · notes
- Dell Precision T7810 (7810), Dual Intel Xeon E5-2650 v4 (2x12c/24t, 2.20GHz), 32GB DDR4 ECC 2133 MT/s (8x4GB), PCIe Gen 3 x16 (8GT/s, both GPUs at x16), both GPUs on same NUMA node (NUMA 0, PHB topology). Layer split (-sm layer), flash attention, f16 KV cache at all contexts. Stock llama.cpp build (commit 13f2b28, NOT PrismML fork). Benchmarked with llama-bench. Dual GPU via llama.cpp layer split. Hybrid SSM/attention: 8 full attention layers (every 4th), 24 SSM/Mamba layers.
- localmaxxing · observed command
- llama-bench -m -p 512 -n 128 -ngl 32 -sm layer -fa on -ctk f16 -ctv f16 -r 5
- localmaxxing · run id
- cmrynblpx042ao401u3m65xb2
- localmaxxing · tokenized · arguments
- llama-bench, -m, -p, 512, -n, 128, -ngl, 32, -sm, layer, -fa, on, -ctk, f16, -ctv, f16, -r, 5
- 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/cmrynblpx042ao401u3m65xb2 ↗ |