Research
Memory
Systems that help AI retain and use context over time.
93.4%
Long-term memory
LongMemEval
467 of 500 questions · four separate judges score it 91.2% to 93.4%
500 questions asked against 40 to 50 prior sessions of conversation, judged against the reference answers.
Backboard-longmemEval-results ↗
90.0%
Conversational memory
LoCoMo
Leads all four categories · +14.2 pts over the next-best system
Every system answers the same questions on the same conversations. GPT-4.1 judges each answer.
Backboard-Locomo-Benchmark ↗
From the blog
- Aug 2026BlogWhen Your AI Remembers: A New Way to Think About Memory in Enterprise AIRead
- Apr 2026BlogThe Hidden Challenge of Multi-LLM Context ManagementRead
- Mar 2026BlogAdaptive Context Window Management Across 17,000+ ModelsRead
- Feb 2026BlogUnderstanding Backboard’s AI Ecosystem: State, RAG, and MemoryRead
- Feb 2026BlogBackboard: 90% on LoCoMo, 91–93% on LongMemEvalRead