LFM2.5 1.2B Instruct on iPhone 12 · 4 GB

Experiment Complete Published

llama.cpp b12110 on iPhone 12 · 4 GB, Metal

LFM2.5 1.2B Instruct as Q8_0 (8-bit), archive 256 input / 128 output: a 256-token prompt and 128 tokens out. Measured 13 Nov 2026, 00:38 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: f16, uzu default. The experiments here use different 8-bit schemes: Q8_0, mirai-l.

Experiment id s4-iphone12-lfm2.5-1.2b-8-bit-llama-256

Summary

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

MetricMedianRangeMean ± SDQuartiles
Decode speed tok/s, higher is better24.1 23.7–24.2 24.0 ± 0.2 24.1–24.1
Time to first token ms, lower is better938 936–998 948 ± 24.7 937–939
Peak memory GiB, lower is betterNot measured
Prefill speed tok/s, higher is better273.1 256.8–273.9 270.6 ± 6.8 273.0–273.4

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
1998256.8256 tok in 997 ms, engine23.7127 tok in 5,348 ms, enginenot recordednominal → nominal
2938273.1256 tok in 937 ms, engine24.2127 tok in 5,252 ms, enginenot recordednominal → nominal
3938273.1256 tok in 937 ms, engine24.0127 tok in 5,282 ms, enginenot recordednominal → nominal
4936273.9256 tok in 935 ms, engine24.1127 tok in 5,269 ms, enginenot recordednominal → nominal
5939273.0256 tok in 938 ms, engine24.1127 tok in 5,270 ms, enginenot recordednominal → nominal
6937273.5256 tok in 936 ms, engine24.1127 tok in 5,264 ms, enginenot recordednominal → 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 iPhone13,2
Cohort
iPhones (mobile)
SoC
Apple A14 Bionic
CPU
6-core CPU (2 performance, 4 efficiency)
GPU
4-core Apple GPU
NPU
16-core Neural Engine
RAM
4.00 GiB
OS
iOS 27.2.1 (24B156)
Kernel
Darwin Kernel Version 27.2.0: Tue Oct 13 21:44:36 PDT 2026; root:xnu-13432.42.3~28/RELEASE_ARM64_T8101
Power
usb
Low power mode
No
App state
active

Engine

Engine
llama.cpp
Maintainer
ggml-org
License
MIT, open source
Version
b12110
Commit
cd959bae99ff4bed22bef2604f45241c09bbb8c8
Binary SHA-256
c1a1252fac9adcbcd54d3ce0f381dba44f6ba05f0561d4b05e79a04228a8ea12
Built from
release
Build flags
None
Backend
Metal
Threads
Not recorded
Interface
ios-app
Settings in effect
backend=metal gpu_layers=999 flash_attn=on cache_type_k=f16 cache_type_v=f16 load_mode=mmap app=XybridBench process=one app launch per generation threads=C API default sampling=greedy prompt=frozen token IDs

Model and weights

Base model
LFM2.5 1.2B Instruct (LFM2.5, 1.2B)
Upstream
LiquidAI/LFM2.5-1.2B-Instruct@b9d816ab510f9a28dca37cb182e9da8b0226e13c
Weights
LiquidAI/LFM2.5-1.2B-Instruct-GGUF@8ed288026e23958ad9dfa92d53ed773a8eee7125
Format
gguf
File 1
LFM2.5-1.2B-Instruct-Q8_0.gguf · 1.16 GiB · sha256 f6b981dcb86917fa463f78a362320bd5e2dc45445df147287eedb85e5a30d26a
File 2
README.md · 3,544 bytes · sha256 f89d69dd107c19936b98ff981456302bc59255cce34b216ecc489d7bafa782e9
Quantization
Q8_0 (8-bit)
Weight bits
8
Group size
32
KV cache
f16
Conversion
Published upstream by LiquidAI
Conversion command
Not recorded

Workload

Workload
Archive 256 input / 128 output
Prompt
256 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
60 s between trials

Conditions

Ambient
Not recorded
Thermal at start
nominal
Thermal at end
nominal
Thermal gate
iOS thermal state nominal
Throttling detected
No

Provenance

Measured by
Glenn Sonna, Xybrid
Operator affiliation
Xybrid
Run at
13 Nov 2026, 00:38 UTC
Harness
xybrid-bench 0.4.0 (e4736b3507296d5d0ad7406a73573b62a402827d)
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
11
Check arithmetic
Passed · sha256 6b51d431df5d7f141cbececcf79edf3dd861c3b4069f0b11661a3eefacbba918
Check capital
Passed · sha256 5dd272b4f316b776a7b8e3d0894b37e1e42be3d5d3b204b8a5836cc50597a6b1
Check opposite
Passed · sha256 f151cd9e12c8994edfa316e4624f956bc68460454869177f287146f8cc956a12
Check weekday
Passed · sha256 7d8af1de1262f150187d5938bf649d6c35726970685f916d1ec6392a801b9762
Check color
Passed · sha256 d486dfbd5fb578340bccbdd0a662527eab38550648d5f44517e5ac71b8824703
Check sort
Passed · sha256 a386b3c9b4b4786df5bb6474bab0a62b8476e2f3f9c8a6433aca40152840f6b7
Check animal
Passed · sha256 4b227777d4dd1fc61c6f884f48641d02b4d121d3fd328cb08b5531fcacdabf8a
Check word
Passed · sha256 3b635d3b4437e023e72a9df26201b6ab51285b376446446ebb379121994550f6
Check copy
Passed · sha256 d61e6890d0e3571b0c7693612d41f8d059a1fcdef4ba959c0af4144541d41907
Check needle-start
Passed · sha256 8b9af2e5b8b4cfddb87d171997b95c471f8d68a68e30fb5177abeb3332372edd
Check needle-end
Passed · sha256 e6589c736af32f81bf239c5eda6de41959122ccc1d05463b75e0eb0c70fc831c
Check continuation-128
Passed · sha256 fe1fdf5d49caa68c1ef2f8a341413cb2c3826023aef1113c6c289dcd86193b32
Detail
12/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none.

Reproduce it

Commands
pnpm bench run --plan plans/s4-iphone12-lfm2.5-1.2b-8-bit.json

Three palindromic blocks per quantization class and workload; load, two warmups, one trial, unload per turn.

Raw data