Back to Arena

Adept AI

by Adept AI Labs (acquired by Amazon 2024)

System Card

OrganizationAdept AI Labs (acquired by Amazon 2024)
Released2022-04
Architectureagentic-workflow / Computer-Use Agent (ACT-1)
DetailsAdept's flagship ACT-1 is a large-scale transformer trained to use digital tools — primarily web browsers and desktop software — by understanding and executing natural language instructions. A proprietary DSL and actuation layer enables actions across arbitrary websites and applications without pre-built integrations. The team was acquired by Amazon in June 2024 to join Reality Labs for agentic AI development.
Parameters—
Domainagent-memorylong-context
Open SourceNo
WebsiteVisit
computer-usebrowser-agentaction-modelamazonacquiredworkflow-automation

Capability Profile

Benchmark Scores

6 of 14 benchmarks
Long-Context Retrieval
2/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
603p
∞Bench
73.519p
Multi-Turn Recall
1/2
LoCoMo
78.680p
MemoryBank
no data
Cross-Session Memory
0/1
LongMemEval
no data
Multi-Hop QA
2/3
BABILong
76.967p
MultiHop-RAG
no data
HotpotQA
76.480p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Adept AI vendor documentation; evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Adept AI vendor documentation; evaluated on BABILong: Testing the Limits of LLMs with Long-Context Reasoning-in-a-Haystack (AIRI, 2406)Adept AI vendor documentation; evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Adept AI vendor documentation; evaluated on InfiniteBench: Extending Long Context Evaluation Beyond 100K Tokens (Tsinghua / OpenBMB, 2402)Adept AI vendor documentation; evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Adept AI vendor documentation; evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)