Recipe
lfm2-5-8b-a1b-q6-k-rtx-3080-10gb-llama-cpp-tp1
lfm2-5-8b-a1b-q6-k-rtx-3080-10gb-llama-cpp-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- llama.cpp
- Engine version
- b10472
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 128,000
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/LiquidAI/LFM2.5-8B-A1B-GGUF- Repository
- LiquidAI/LFM2.5-8B-A1B-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-mLFM2.5-8B-A1B-Q6_K.gguf--host127.0.0.1--port8099--deviceCUDA0--n-gpu-layers99--ctx-size128000--flash-attnon--cache-type-kq8_0--cache-type-vq8_0--parallel1--jinja--spec-typenone
| Flag | Value |
|---|---|
-m | LFM2.5-8B-A1B-Q6_K.gguf |
--host | 127.0.0.1 |
--port | 8099 |
--device | CUDA0 |
--n-gpu-layers | 99 |
--ctx-size | 128000 |
--flash-attn | on |
--cache-type-k | q8_0 |
--cache-type-v | q8_0 |
--parallel | 1 |
--spec-type | none |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 128,000 | 9,840.7 | 277.8 | 286.4 | observed | lfm2-5-8b-a1b-q6-k-rtx-3080-10gb-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-3080-10gb
- id
- lfm2-5-8b-a1b-q6-k-rtx-3080-10gb-llama-cpp-tp1
- model instance id
- liquidai-lfm2-5-8b-a1b-gguf--q6-k
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- lfm2-5-8b-a1b-q6-k-rtx-3080-10gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
draft launch
- accelerator backend
- nvidia
- arguments
- -hf, LiquidAI/LFM2.5-8B-A1B-GGUF:Q6_K, --n-gpu-layers, 999, --host, 0.0.0.0, --port, 8080, -c, 128000
- 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
- version
- b10472
serving
- max concurrency
- 1
- max context tokens
- 128,000
- 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 · hardware label
- RTX 3080
- localmaxxing · notes
- Companion to the Q4_K_M entry on the same card, for the quant-vs-speed curve: Q6_K gives 278 tok/s in 8266 MiB versus Q4_K_M at 328 tok/s in 6546 MiB, both at 128K context and both fully resident. Sparse MoE, 8B total with ~1B active per token, which is why decode is far above a dense 8B. Measured on an RTX 3080 10GB. Decode is the MEDIAN of 3 reps, not the best. Prompt is deliberately non-repetitive (repeated filler inflates decode on speculative setups by ~25%) and ignore_eos is set so output token count is identical across runs. Peak VRAM 8266 MiB with 212 MiB host-shared; elevated host-shared only indicates spill when the decode rate drops with it.
- localmaxxing · observed command
- llama-server \ -m LFM2.5-8B-A1B-Q6_K.gguf \ --host 127.0.0.1 --port 8099 \ --device CUDA0 --n-gpu-layers 99 \ --ctx-size 128000 \ --flash-attn on \ --cache-type-k q8_0 --cache-type-v q8_0 \ --parallel 1 --jinja \ --spec-type none
- localmaxxing · run id
- cmt1zgkzp08yxmv01h4zcha9k
- localmaxxing · tokenized · arguments
- llama-server, -m, LFM2.5-8B-A1B-Q6_K.gguf, --host, 127.0.0.1, --port, 8099, --device, CUDA0, --n-gpu-layers, 99, --ctx-size, 128000, --flash-attn, on, --cache-type-k, q8_0, --cache-type-v, q8_0, --parallel, 1, --jinja, --spec-type, none
- 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/cmt1zgkzp08yxmv01h4zcha9k ↗ |