system-one-bench

~/system-one-bench / models/liquid-ai-d1-1

Liquid AI · OpenRouter API

d1.1

d1.1 by Liquid AI scores 82.3 on the System One Index (10th of 39) and 85.8 on the Vision Index (3rd of 14). Calibration error 2.1%.

System One Index
82.3
10th of 39
English
81.3
Arabic
83.3
Vision Index
85.8
3rd of 14
Calibration error
2.1%
Lower is better
Median latency
–
Cost per 1,000 calls
$0.005
As billed

Text datasets

DatasetAccuracyRankBest on the boardCalibration error
AG News English90.510th of 3992.55.6%
SST-2 English98.01st of 3998.06.4%
SST-5 English60.07th of 3962.54.5%
DAIR Emotion English60.512th of 3991.511.9%
Banking77 English87.019th of 3998.57.1%
BoolQ English92.03rd of 3993.02.6%
XNLI English English90.014th of 3995.35.9%
Typed decisions English72.018th of 3980.05.9%
XNLI French French83.311th of 3991.37.8%
XNLI Arabic Arabic79.39th of 4190.08.2%
MASSIVE scenario Arabic73.011th of 4189.510.0%
MASSIVE intent Arabic88.011th of 4191.54.9%
OSACT4 offensive Arabic92.511th of 4197.07.8%
Tweet sentiment Arabic72.59th of 4180.04.4%
MIRACL rerank Arabic88.59th of 4190.52.2%
MIRACL pairwise Arabic89.517th of 4191.53.8%

Image datasets

DatasetAccuracyRankBest on the boardCalibration error
SST-2, as an image96.52nd of 1497.05.6%
Arabic sentiment, as an image61.06th of 1475.05.7%
Oxford-IIIT Pet90.53rd of 1498.06.7%
VQAv2 yes/no89.51st of 1489.53.2%
CountBench91.57th of 1495.56.9%

This run

Model id
liquid/d1.1
Runs on
OpenRouter API
Run date
2026-10-30
Latency
API; not timed (round 9 ran from a machine whose round trips don't compare with the board's)
Questions
4,850, 0 failed (counted wrong)

The successor to d1, served as liquid/d1.1-20261027. 64K context. Doesn't answer Box questions.

Compare

Ranks count the models charted on the leaderboard. Every model gets the same cases and the same request, under the benchmark rules.