researchAgent Frameworks4d ago
Folio: A New Context Management System for Long-Horizon AI Agents
Facts
Folio is a novel context manager designed to improve performance for long-horizon AI agents by using multi-resolution history folding. By managing history through linked levels instead of model-based summarization, it reduces token costs and execution time while improving task completion rates on benchmarks.
- ▸Folio uses a three-level (L1, L2, L3) linked structure to store and fold agent history without relying on model-written summaries.
- ▸The system achieved a 66/75 success rate on LOCA-bench tasks using DeepSeek-V4-Pro.
- ▸Folio outperformed standard model-based compaction (54/75) and other solutions like Claude Code, Codex, and SoL-Pi (49-59/75).
- ▸The system reportedly operates at 2.4 times lower cost and completes tasks in half the working time compared to baseline methods.
- ▸Each fold operation takes a median of 0.6 seconds.
- Occurred
- 2026-11-12 10:59 UTC
- First reported
- 2026-11-12 10:59 UTC
- Last updated
- 2026-11-13 06:16 UTC
- Event type
- research
- Region
- global
Entities
Categories
Source timeline · 1
- arXivFirst-party2026-11-12 10:59
Community vote
Sign in to vote on importance and sentiment.
Platform Signals
Impact score
70
AI sentiment
PositiveEvent significance30%1
Entity importance20%1
Source credibility20%1
Novelty15%1
Corroboration15%1
Confidence1
Novelty1
Corroboration1
Community
ImportanceNo votes
Strategic sentiment
Positive 0%Neutral 0%Negative 0%
Platform signals are algorithmic and community-derived indicators. They are not investment recommendations and do not predict valuation or performance.
Related events
- 3dSurveyWeave Agentic Framework for Automated Survey Generation
- 4dIntroduction of Singularity-Bench for Evaluating Autonomous AI Development
- 5dResearchers Introduce WordInferBench to Evaluate AI Agent Contextual Inference in Document Editing
- 6dOpenAI Demonstrates Advanced Developer Workflows for Codex at AI Engineer Code Summit
- 6dIntroduction of RSI-PDEbench and RSI-PDE Agent Framework for Scientific Computing
- 6dResend Discusses Designing APIs for AI Agent Consumption