Name: Agnes 3.0 Flash (hosted)
Author: Sapiens AI

Agnes 3.0 Flash (hosted)

Sapiens AI

Sapiens AI's proprietary production model evaluated by Artificial Analysis. The publisher explicitly distinguishes it from the 33B Agnes-3.0-Flash Preview weights on Hugging Face; their scores and architecture are not interchangeable.

Benchmark scores

Benchmark	Score
AA-LCR v1.1	81.0
AA-Omniscience Index	-10.6
Artificial Analysis Intelligence Index v4.3	35.5
Artificial Analysis Intelligence Index v4.3.2	35.5
Artificial Analysis Omniscience Accuracy	25.5
Artificial Analysis Omniscience Hallucination Rate	48.4
AutomationBench-AA 1.0.6	50.7
Capability	158.4
CritPt (avg@5)	15.1
GDP.pdf (Artificial Analysis, All-pass)	10.0
GDP.pdf (Artificial Analysis, Mean Criteria)	40.0
GDPval-AA v2	1573
GDPval-AA v2.1	1550
GPQA Diamond (Artificial Analysis avg@5)	92.4
Humanity's Last Exam (Text-Only)	38.5
SciCode 1.0.1 (Artificial Analysis)	51.6
Terminal-Bench 2.1	82.0
Terminal-Bench 4.0 (Artificial Analysis)	7.1
𝜏³-Banking	47.6