Recipe
gemma-4-12b-q4-k-xl-rtx-3060-12gb-llama-cpp-tp1
gemma-4-12b-q4-k-xl-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
- Engine version
- b9775 / be4a6a63e
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 4,096
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/google/gemma-4-12B-it- Repository
- google/gemma-4-12B-it
- 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-mgemma-4-12B-it-qat-UD-Q4_K_XL.gguf--model-draftgemma-4-12B-it-qat-assistant-MTP-Q8_0.gguf--spec-typedraft-mtp--spec-draft-n-max3-ngl99-faon-ctkq4_0-ctvq4_0-c196608-np1
| Flag | Value |
|---|---|
-m | gemma-4-12B-it-qat-UD-Q4_K_XL.gguf |
--model-draft | gemma-4-12B-it-qat-assistant-MTP-Q8_0.gguf |
--spec-type | draft-mtp |
--spec-draft-n-max | 3 |
-ngl | 99 |
-fa | on |
-ctk | q4_0 |
-ctv | q4_0 |
-c | 196608 |
-np | 1 |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 4,096 | 114.9 | 89.7 | — | observed | gemma-4-12b-q4-k-xl-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
- gemma-4-12b-q4-k-xl-rtx-3060-12gb-llama-cpp-tp1
- model instance id
- google-gemma-4-12b-it--q4-k-xl
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- gemma-4-12b-q4-k-xl-rtx-3060-12gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- b9775 / be4a6a63e
serving
- max concurrency
- 1
- max context tokens
- 4,096
- 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
- Gemma4 12B QAT Q4_K_XL + MTP draft head (Q8_0, 444 MB). OPTIMAL: MTP enabled (--spec-type draft-mtp --spec-draft-n-max 3, draft model gemma-4-12B-it-qat-assistant-MTP-Q8_0.gguf). 42 dense attn layers, fully offloaded. KV cache q4_0. MTP speedup: 2.38x. Acceptance rate: 94% (188/200). Max context with MTP: 196608 (192K) at ~11.5 GB. Fedora 43, CUDA 13.0.
- localmaxxing · observed command
- llama-server -m gemma-4-12B-it-qat-UD-Q4_K_XL.gguf --model-draft gemma-4-12B-it-qat-assistant-MTP-Q8_0.gguf --spec-type draft-mtp --spec-draft-n-max 3 -ngl 99 -fa on -ctk q4_0 -ctv q4_0 -c 196608 -np 1
- localmaxxing · run id
- cmr3ynprv0007nv0192k93bri
- localmaxxing · tokenized · arguments
- llama-server, -m, gemma-4-12B-it-qat-UD-Q4_K_XL.gguf, --model-draft, gemma-4-12B-it-qat-assistant-MTP-Q8_0.gguf, --spec-type, draft-mtp, --spec-draft-n-max, 3, -ngl, 99, -fa, on, -ctk, q4_0, -ctv, q4_0, -c, 196608, -np, 1
- 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/cmr3ynprv0007nv0192k93bri ↗ |