Gate News message, April 11, AI infrastructure company Ramp Labs released research findings called “Latent Briefing,” enabling efficient memory sharing among multi-agent systems by directly compressing large-model KV caches, greatly reducing Token consumption without losing accuracy. In mainstream multi-agent architectures, the orchestrator breaks down tasks and repeatedly calls worker model instances; as the inference chain grows longer, Token usage expands exponentially. The core idea behind Latent Briefing is to use the attention mechanism to identify the truly crucial parts of the context, discard redundant information directly at the representation layer, rather than relying on slow LLM summarization or RAG retrieval with less stable results. On the LongBench v2 benchmark, the method performed impressively: the worker model’s Token consumption dropped by 65%, the Token savings’ median for medium-length documents (32k to 100k) reached 49%, overall accuracy improved by about 3 percentage points versus the baseline, and the additional time spent per compression was only about 1.7 seconds—roughly a 20x speedup compared with the original algorithm. The experiments used Claude Sonnet 4 as the orchestrator and Qwen3-14B as the worker model, covering a wide range of document scenarios including academic papers, legal documents, novels, and government reports. The study also found that the optimal compression threshold varies with task difficulty and document length—hard problems are better suited to aggressive compression to filter speculative reasoning noise, while long documents are better suited to lighter compression to preserve dispersed key information.
Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to
Disclaimer.
Related Articles
OpenClaw Releases v2026.4.29 on April 29, Upgrades Memory to Personalized Wiki with Relationship Tracking
According to Beating, open-source AI assistant OpenClaw (GitHub 367K stars) released v2026.4.29 on April 29, marking its second update in two days. The memory system evolved from simple retrieval-based recall to personalized wiki, enabling agents to automatically build character profiles and track r
GateNews3h ago
Google CEO Pichai reveals that using Gemini AI to understand human nature helps build more sincere communication
Pichai said that before important meetings, he uses Gemini’s perspective to analyze and predict the other party’s psychology, thereby improving empathy and enabling more sincere communication. AI agents can also automatically organize emails, scheduling, and summaries, making everyday chores more efficient. Meanwhile, AI platforms centered on open co-creation are emerging; open-source technologies such as Gemini 4 lower the barrier to entry. At the same time, it emphasizes building AI governance frameworks, with governments and society needing to participate to address challenges such as cybersecurity, deepfakes, and sustainability.
ChainNewsAbmedia5h ago
Oobit Launches Visa-Supported AI Agent Cards on Thursday, Enabling USDT Spending Without Fiat Conversion
According to The Block, Tether-backed wallet startup Oobit launched AI Agent Cards on Thursday, allowing autonomous bots to make purchases using USDT balances without converting to fiat or accessing corporate card credentials directly. The Visa-supported cards are usable online wherever Visa is acce
GateNews6h ago
ChimpX AI Raises $2.8M in Seed Round Led by Waterdrip Capital and MetaLabs Ventures
ChimpX AI announced today the close of a $2.8 million seed round to accelerate development of Mojo AI, an execution agent that converts plain-English intent into on-chain DeFi transactions on Solana. The round was led by Waterdrip
GateNews8h ago
Major CEX Launches Agent Payments Protocol on April 29, Enabling AI-Driven Cross-Chain Transactions
According to a recent announcement, a leading cryptocurrency exchange unveiled the Agent Payments Protocol on April 29, an open standard enabling artificial intelligence agents to execute full business transactions across multiple blockchain networks without human intervention. The protocol
GateNews8h ago
Walrus Launches MemWal SDK for AI Agent Memory
Walrus has launched MemWal, an SDK designed to address limitations in agentic memory by bringing verifiability, availability, portability and sharability to how AI agents store and access information, according to Mysten Labs Group Product Manager Abinhav Garg.
Verifiable and Portable Memory
CryptoFrontier9h ago
