LFM2.5 350M on M5 Max · 64 GiB

Experiment Complete Published

LiteRT-LM 0.19.1 on M5 Max · 64 GiB, CPU

LFM2.5 350M as litert-int8 (8-bit), archive 2048 input / 128 output: a 2048-token prompt and 128 tokens out. Measured 12 Nov 2026, 20:25 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: litert-int8, where most use Q8_0.

Experiment id s3-lfm2.5-350m-8-bit-litert-2048

Summary

Over 6 valid trials of 6 attempted (6 planned), computed from the raw trials below.

MetricMedianRangeMean ± SDQuartiles
Decode speed tok/s, higher is better198.9 198.3–200.6 199.2 ± 1.0 198.4–200.1
Time to first token ms, lower is better635 631–643 636 ± 4.6 632–638
Peak memory GiB, lower is better1.25 1.25–1.25 1.25 ± 0.00 1.25–1.25
Prefill speed tok/s, higher is better3,308 3,268–3,330 3,306 ± 23.8 3,296–3,325

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.

Every attempted trial
TrialTime to first token msPrefill tok/sDecode tok/sPeak memory GiBThermalNote
16333,3192,048 tok in 617 ms, engine200.4127 tok in 634 ms, engine1.25nominal → nominal
26313,3302,048 tok in 615 ms, engine198.5127 tok in 640 ms, engine1.25nominal → nominal
36323,3272,048 tok in 615 ms, engine198.4127 tok in 640 ms, engine1.25nominal → nominal
46383,2952,048 tok in 622 ms, engine200.6127 tok in 633 ms, engine1.25nominal → nominal
56433,2682,048 tok in 627 ms, engine199.2127 tok in 638 ms, engine1.25nominal → nominal
66393,2972,048 tok in 621 ms, engine198.3127 tok in 641 ms, engine1.25nominal → 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
LiteRT-LM
Maintainer
Google AI Edge
License
Apache-2.0, open source
Version
0.19.1
Commit
Not recorded
Binary SHA-256
c61e0faa71026f2875a710a42cbe803226004db811f7ee47ea78198a0499e851
Built from
package
Build flags
None
Backend
CPU
Threads
Not recorded
Interface
python-worker
Settings in effect
backend=cpu max_num_tokens=4096 enable_benchmark=true chat_template=pass-through process=one per generation sampler=top_k=1 temperature=0 stop_token_ids=[[7], [2]] cache_dir=[runner]/engines/litert-lm/cache bos_token_id=1

Model and weights

Base model
LFM2.5 350M (LFM2.5, 350M)
Upstream
LiquidAI/LFM2.5-350M@9921cd4d82c0e3986887aa4b376f9446729ef8a3
Weights
litert-community/LFM2.5-350M@4754bf697659039afee2962cfd9c8f40d547ab35
Format
litertlm
File 1
LFM2.5-350M_int8.litertlm · 384.8 MiB · sha256 a6b191d201e96b876d14eefb9bc34b7da05cc3951e28d06c3ad252481b36c121
File 2
README.md · 14,806 bytes · sha256 4aa8f3ef769b89ecf207b1506da4a8c617cd7aafbcf9520a96ae2e10f1a6bc56
Quantization
litert-int8 (8-bit)
Weight bits
8
Group size
Not recorded
KV cache
litert-lm default
Conversion
Published upstream by litert-community
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, 20:25 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

Commands
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