Recipe
phi-4-mini-instruct-gguf-q8-0-intel-arc-pro-b70-32gb-llama-cpp-tp1
phi-4-mini-instruct-gguf-q8-0-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
- fdb1db877
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 4,096
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF- Repository
- bartowski/microsoft_Phi-4-mini-instruct-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-m/mnt/usb-models/llm-models/phi-4-mini-instruct-gguf/microsoft_Phi-4-mini-instruct-Q8_0.gguf-c4096-ngl99-b1024-ub256-faon--ctx-checkpoints0--jinja--reasoningoff
| Flag | Value |
|---|---|
-m | /mnt/usb-models/llm-models/phi-4-mini-instruct-gguf/microsoft_Phi-4-mini-instruct-Q8_0.gguf |
-c | 4096 |
-ngl | 99 |
-b | 1024 |
-ub | 256 |
-fa | on |
--ctx-checkpoints | 0 |
--reasoning | off |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 4,096 | — | 72.2 | 119.5 | observed | phi-4-mini-instruct-gguf-q8-0-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
- phi-4-mini-instruct-gguf-q8-0-intel-arc-pro-b70-32gb-llama-cpp-tp1
- model instance id
- bartowski-microsoft-phi-4-mini-instruct-gguf--gguf-q8-0
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- phi-4-mini-instruct-gguf-q8-0-intel-arc-pro-b70-32gb-llama-cpp-tp1-sweep
- status
- candidate
capabilities
engine
- name
- llama.cpp
- version
- fdb1db877
serving
- 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
- serving.max concurrency · provenance · captured at
- 2026-08-31T23:03:15Z
serving.max concurrency · reason server-capacity-not-evidenced
serving.max concurrency · state unknown
metadata
- localmaxxing · backend
- xpu
- localmaxxing · batch size
- 1
- localmaxxing · hardware label
- Intel Arc Pro B70
- localmaxxing · notes
- Strict fresh-response rapid snapshot for Phi-4-mini-instruct Q8_0 on one Intel Arc Pro B70. Fixed rapid realistic suite, one cold response per prompt, cached_tokens=0 on every row, no prompt/KV/context/response reuse, no n-gram/history acceleration, no speculation. llama.cpp/SYCL, ctx=4096, batch=1024, ubatch=256, FlashAttention on, f16 KV, server --cache-ram 0 and per-request cache_prompt=false. Standalone confirmations were 72.884 and 72.246 tok/s; the lower repeat is promoted conservatively. This is a higher-quality quant reference, not the fastest Phi-4-mini row.
- localmaxxing · observed command
- llama-server -m /mnt/usb-models/llm-models/phi-4-mini-instruct-gguf/microsoft_Phi-4-mini-instruct-Q8_0.gguf -c 4096 -ngl 99 -b 1024 -ub 256 -fa on --ctx-checkpoints 0 --jinja --reasoning off
- localmaxxing · run id
- cmr6yazvy00hgmn01s5rtowwa
- localmaxxing · tokenized · arguments
- llama-server, -m, /mnt/usb-models/llm-models/phi-4-mini-instruct-gguf/microsoft_Phi-4-mini-instruct-Q8_0.gguf, -c, 4096, -ngl, 99, -b, 1024, -ub, 256, -fa, on, --ctx-checkpoints, 0, --jinja, --reasoning, 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/cmr6yazvy00hgmn01s5rtowwa ↗ |