Recipe
qwen3-5-4b-mlx-8bit-apple-m4-pro-48gb-lmstudio-tp1
qwen3-5-4b-mlx-8bit-apple-m4-pro-48gb-lmstudio-tp1Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.
Record
- Status
- candidate
- Source
- localmaxxing
- Engine
- lmstudio
- Engine version
- 0.4.20+1
- Accelerators
- 1
- Tensor parallel
- 1
- Context tokens
- 262,144
- Max concurrency
- 1
- chat
- unknown
- reasoning
- unknown
- tools
- unknown
- vision
- unknown
Hugging Face model card
Identity
https://huggingface.co/mlx-community/Qwen3.5-4B-MLX-8bit- Repository
- mlx-community/Qwen3.5-4B-MLX-8bit
- 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
lmstudio
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.
lmsserverstart--port1234--bind127.0.0.1
lmsloadqwen3.5-4b-mlx--context-length262144--parallel4--identifierbenchmark-qwen35-4b-mlx--yes
| Flag | Value |
|---|---|
--context-length | 262144 |
--parallel | 4 |
--identifier | benchmark-qwen35-4b-mlx |
Measured speed
| Concurrency | Context | Prefill | Decode | TTFT ms | Status | Sweep |
|---|---|---|---|---|---|---|
| 1 | 262,144 | 743.2 | 49 | 736 | observed | qwen3-5-4b-mlx-8bit-apple-m4-pro-48gb-lmstudio-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
- apple-m4-pro-48gb
- id
- qwen3-5-4b-mlx-8bit-apple-m4-pro-48gb-lmstudio-tp1
- model instance id
- mlx-community-qwen3-5-4b-mlx-8bit--mlx-8bit
- recipe source
- localmaxxing
- schema version
- local-ai-registry/v1
- speed sweep ids
- qwen3-5-4b-mlx-8bit-apple-m4-pro-48gb-lmstudio-tp1-sweep
- status
- candidate
capabilities
engine
- name
- lmstudio
- version
- 0.4.20+1
serving
- max concurrency
- 1
- max context tokens
- 262,144
- 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
- metal
- localmaxxing · hardware label
- M4 Pro
- localmaxxing · notes
- MLX 8-bit quantization. Median of 3 timed runs after 1 warmup using the same deterministic 512-word source-prompt protocol at temperature 0. The model-specific chat template produced 547 prompt tokens; each run generated 254 output tokens. Prefill estimated as prompt tokens divided by LM Studio native time-to-first-token.
- localmaxxing · observed command
- lms server start --port 1234 --bind 127.0.0.1 && lms load qwen3.5-4b-mlx --context-length 262144 --parallel 4 --identifier benchmark-qwen35-4b-mlx --yes
- localmaxxing · run id
- cmsqri9h200qvmr01wcj8yjo5
- localmaxxing · tokenized · arguments
- lms, load, qwen3.5-4b-mlx, --context-length, 262144, --parallel, 4, --identifier, benchmark-qwen35-4b-mlx, --yes
- localmaxxing · tokenized · fidelity
- faithful
localmaxxing · tokenized · steps
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
|---|---|---|---|---|---|---|---|
| lms | server | start | --port | 1234 | --bind | 127.0.0.1 | Unknown |
| lms | load | qwen3.5-4b-mlx | --context-length | 262144 | --parallel | 4 | --identifier |
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/cmsqri9h200qvmr01wcj8yjo5 ↗ |