Experiment Complete Published
MLX LM 0.32.2 on M5 Max · 64 GiB, Metal
LFM2.5 350M as mlx-affine-8bit-g64 (8-bit), archive 2048 input / 128 output: a 2048-token prompt and 128 tokens out. Measured 12 Nov 2026, 20:24 UTC by Glenn Sonna, Xybrid.
Eligible for ranking when the objective's required metrics are measured and this configuration is selected.Approximate comparison: The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
Experiment id s3-lfm2.5-350m-8-bit-mlx-2048
Summary
Over 6 valid trials of 6 attempted (6 planned), computed from the raw trials below.
| Metric | Median | Range | Mean ± SD | Quartiles |
|---|---|---|---|---|
| Decode speed tok/s, higher is better | 702.2 | 695.5–703.1 | 701.0 ± 2.8 | 701.1–702.2 |
| Time to first token ms, lower is better | 57.7 | 57.2–57.9 | 57.6 ± 0.3 | 57.5–57.9 |
| Peak memory GiB, lower is better | 2.09 | 1.94–2.38 | 2.10 ± 0.16 | 1.98–2.16 |
| Prefill speed tok/s, higher is better | 35,880 | 35,780–36,186 | 35,931 ± 169.3 | 35,793–36,044 |
Trials, metric by metric
The line runs from the lowest to the highest valid trial, the dot is the median and each tick is one trial.
Every trial
Raw token counts and durations, with the speeds computed from them. Invalid trials stay in the record with their reason; they are left out of the summary, never silently dropped.
| Trial | Time to first token ms | Prefill tok/s | Decode tok/s | Peak memory GiB | Thermal | Note |
|---|---|---|---|---|---|---|
| 1 | 57.9 | 35,7802,048 tok in 57.2 ms, engine | 702.1127 tok in 181 ms, engine | 1.94 | nominal → nominal | |
| 2 | 57.2 | 36,1862,048 tok in 56.6 ms, engine | 703.1127 tok in 181 ms, engine | 2.12 | nominal → nominal | |
| 3 | 57.9 | 35,7822,048 tok in 57.2 ms, engine | 702.2127 tok in 181 ms, engine | 2.38 | nominal → nominal | |
| 4 | 57.8 | 35,8252,048 tok in 57.2 ms, engine | 702.2127 tok in 181 ms, engine | 2.18 | nominal → nominal | |
| 5 | 57.7 | 35,9352,048 tok in 57.0 ms, engine | 700.8127 tok in 181 ms, engine | 2.05 | nominal → nominal | |
| 6 | 57.4 | 36,0802,048 tok in 56.8 ms, engine | 695.5127 tok in 183 ms, engine | 1.95 | nominal → nominal |
Caveats
- Publisher cards identify the base model but do not always pin the exact export source checkpoint revision.
- Engines may differ in quantization scheme, effective KV cache precision, kernels, and activations.
Configuration
Device
- Device
- Apple Mac17,6
- Cohort
- Apple silicon Macs (desktop)
- SoC
- Apple M5 Max
- CPU
- 18-core Apple CPU
- GPU
- Not recorded
- NPU
- Not recorded
- RAM
- 64.00 GiB
- OS
- macOS 27.2.0 (26B123)
- Kernel
- 27.2.0
- Power
- mains
- Low power mode
- No
- Performance mode
- Not recorded
Engine
- Engine
- MLX LM
- Maintainer
- Apple (ml-explore)
- License
- MIT, open source
- Version
- 0.32.2
- Commit
- Not recorded
- Binary SHA-256
- 9d46803799f99d6fc7c35fff2d4357c79b1a25ae98cc63efeaaebeb50a7b5f33
- Built from
- package
- Build flags
- None
- Backend
- Metal
- Threads
- Not recorded
- Interface
- python-worker
- Settings in effect
- backend=Device(gpu, 0) prompt_cache=false speculative=false greedy=true temperature=0 sampler=argmax prefill_step_size=2048 context_length=4096 kv_quantized=false batch_size=1 trust_remote_code=false weight_bits=8 weight_group_size=64
Model and weights
- Base model
- LFM2.5 350M (LFM2.5, 350M)
- Upstream
- LiquidAI/LFM2.5-350M@9921cd4d82c0e3986887aa4b376f9446729ef8a3
- Weights
- LiquidAI/LFM2.5-350M-MLX-8bit@2eb1bb33c541750066fad9757713ff555b3ca4bb
- Format
- safetensors
- File 1
- README.md · 2,329 bytes · sha256 85f447c49d57a798d8270bb5d494786961188f05c4abbb11c98ad0c3b0124244
- File 2
- chat_template.jinja · 2,607 bytes · sha256 1a2d8281b5477abb5bb1cebd69d7d1ee389e074656ca5f5e0ef6ae56d6ae65c1
- File 3
- config.json · 1,714 bytes · sha256 3e5f60543d3712d8431235a95fb6b23d0d8c88708065d1fd966eeb36f8d4a775
- File 4
- generation_config.json · 132 bytes · sha256 1c72d35bb46c5102537eb114fea9a5c10ef928cc813401c3c12aef399c1f8e5d
- File 5
- model.safetensors · 380.5 MiB · sha256 5c77c877b92370344e16a90c0d4bb8bccfad3f0a14f4249ae8db7e7af761822e
- File 6
- model.safetensors.index.json · 23,413 bytes · sha256 14877d5294bf25cc3565e19236d50ee2fe60ce3048971a01fb78b915be096850
- File 7
- tokenizer.json · 4.5 MiB · sha256 df1d8d5ec5d091b460562ffd545e4a5e91d17d4a0db7ebe733be34ed374377bd
- File 8
- tokenizer_config.json · 489 bytes · sha256 75c287923e252b08b0a0f1c367bbe557ab23a681d0b71c5a34e0932ddbe2f5ee
- Quantization
- mlx-affine-8bit-g64 (8-bit)
- Weight bits
- 8
- Group size
- 64
- KV cache
- model-dtype
- Conversion
- Published upstream by LiquidAI
- Conversion command
- Not recorded
Workload
- Workload
- Archive 2048 input / 128 output
- Prompt
- 2,048 tokens, from prompts/source.txt
- Prompt SHA-256
- c10d1b7bef62d95322f8f8d229dc66118d9375107bdc49b06f76287d058b2f6c
- Chat template
- ba551d58630afa3190b1be3602e28301f3d2e9bbac978dfc49d6d825171648b6
- Output
- 128 tokens, ignoring end of sequence
- Context length
- 4,096
- Batch size
- 1
- Sampling
- Greedy
- Timing
- One streamed request per trial
- Warmup
- 2 runs, not counted
- Trials planned
- 6
- Cooldown
- 15 s between trials
Conditions
- Ambient
- Not recorded
- Thermal at start
- nominal
- Thermal at end
- nominal
- Thermal gate
- macOS thermal state nominal
- Throttling detected
- No
Provenance
- Measured by
- Glenn Sonna, Xybrid
- Operator affiliation
- Xybrid
- Run at
- 12 Nov 2026, 20:24 UTC
- Harness
- xybrid-bench 0.4.0 (e16a0a33bb42c49f9e719fb636d4a390e66babe6)
- Methodology
- v0.6
- Publication
- published
- Published
- 16 Nov 2026, 10:23 UTC
- Reviewed by
- Glenn Sonna
Output check
- Result
- Passed
- Method
- sanity-v1-bf16-reference
- Reference
- checks/sanity-v1.json
- Reference SHA-256
- a702dee079525ae0ec90465b1c9e26052c99cb3e1c53e83dea05be4c194fc7ca
- Required passing checks
- 7
- Check arithmetic
- Failed · sha256 4a44dc15364204a80fe80e9039455cc1608281820fe2b24f1e5233ade6af1dd5
- Check capital
- Passed · sha256 5dd272b4f316b776a7b8e3d0894b37e1e42be3d5d3b204b8a5836cc50597a6b1
- Check opposite
- Passed · sha256 f151cd9e12c8994edfa316e4624f956bc68460454869177f287146f8cc956a12
- Check weekday
- Passed · sha256 7d8af1de1262f150187d5938bf649d6c35726970685f916d1ec6392a801b9762
- Check color
- Passed · sha256 d486dfbd5fb578340bccbdd0a662527eab38550648d5f44517e5ac71b8824703
- Check sort
- Failed · sha256 edb794d615a63f5eec436d83bed7299d6af36f508c0bf38a3c1cd85d6e4ebd56
- Check animal
- Passed · sha256 4b227777d4dd1fc61c6f884f48641d02b4d121d3fd328cb08b5531fcacdabf8a
- Check word
- Failed · sha256 df7e70e5021544f4834bbee64a9e3789febc4be81470df629cad6ddb03320a5c
- Check copy
- Passed · sha256 d61e6890d0e3571b0c7693612d41f8d059a1fcdef4ba959c0af4144541d41907
- Check needle-start
- Passed · sha256 8b9af2e5b8b4cfddb87d171997b95c471f8d68a68e30fb5177abeb3332372edd
- Check needle-end
- Passed · sha256 e6589c736af32f81bf239c5eda6de41959122ccc1d05463b75e0eb0c70fc831c
- Check continuation-128
- Failed · sha256 d7f917cdf362c04c176faac2a3a39c878a91838b9152de3ff5b965c02ccc8763
- Detail
- 8/8 of the checks the model passes at BF16; needs 7 and the mandatory ones (needle-start, needle-end). Left out, failed by every BF16 configuration: arithmetic, sort, word, continuation-128.
Reproduce it
pnpm bench run --plan plans/s3-lfm2.5-350m-8-bit.json
Three palindromic blocks per quantization class and workload; load, two warmups, one trial, unload per turn.
Raw data
- This record, as committed JSON, schema v0.3.0
- Complete finalized session evidence raw ·
sha256 552c273db2a0…