The paper introduces Coca-Cola, a continuous latent context compression model that transforms raw tool outputs into latent tokens for LLM agents, bypassing the need for explicit text representation. By embedding these latent tokens into the agent’s input space and employing Action‑Preserving Training, the method ensures that decision‑critical information is retained while reducing token consumption by 70.01%–89.24%. Experiments on web‑search agents across multiple QA benchmarks demonstrate significant reductions in runtime (24.45%–55.97%) and consistent improvements in task performance.
arXiv:2510. 00615v3 Announce Type: replace Abstract: Large language models (LLMs) are increasingly deployed as agents in dynamic real-world environments, where success depends on maintaining precise records of actions and observations.
The paper introduces LOHA, a context layout that compresses older tool observations into soft tokens while keeping the agent’s own turns and the last K observations in plain text, and ACD, a training method that distills full‑text predictions into this latent representation while anchoring behavior on plain text. This approach reduces context per call by up to 57% without significant loss in resolve rates, and improves instance throughput in single‑GPU serving. Experiments on SWE‑bench Verified show that K=3 yields a 43–57% compression with only modest performance impact, while larger windows favor task performance over compression.
arXiv:2606. 31564v1 Announce Type: new Abstract: The increasing complexity of agentic tasks has led to rapidly growing trajectory lengths, which poses significant challenges for large language model (LLM) based agents with fixed context windows.
The paper introduces Agentic Soft-Token Compression (ASTC), a training‑free method that uses soft tokens within an LLM agent to compress long‑horizon context. ASTC leverages the model’s recurrent dynamics and Gated Delta Network attention to identify absorbed information, forwarding it as soft tokens while keeping hard‑token values intact. Experiments show ASTC matches or surpasses uncompressed agents, reducing attention pairs by up to 50.68% and outperforming trained compressors by 15.4 to 103.7 times in attention usage.
The paper introduces FOCUS, a training‑free framework that compresses the interaction history of large language model agents by preserving only the past interactions that causally influence future decisions. Unlike prior methods that learn compression policies offline, FOCUS operates entirely at test time, requiring no additional data collection or fine‑tuning and can be applied to any closed‑API model. Experiments on a variety of agentic benchmarks show that FOCUS reduces peak context length by up to 48% and dependency by 73%, while improving task success by up to 8.9 percentage points.
arXiv:2511. 20639v3 Announce Type: replace-cross Abstract: Multi-agent systems (MAS) extend large language models (LLMs) from independent single-model reasoning to coordinative system-level intelligence.
StateComp introduces a method for long‑horizon agents to decide when to compress historical interactions based on the current agent state, rather than relying on fixed windows or periodic schedules. The framework uses a two‑stage annotation process to create KEEP and READY labels, trains an imbalance‑aware router on frozen language model representations, and groups adjacent READY interactions into compact summaries. Experiments on WorkBuddyBench show that StateComp cuts agent and summarization tokens by 52.27% and speeds up representation extraction 12.67‑fold while preserving task performance.
LLM-based multi-agent systems (MAS) increasingly use latent collaboration to avoid the information loss and repeated encoding-decoding overhead of natural-language communication. However, directly for...
arXiv:2609.37017v1 Announce Type: new Abstract: LLM-based multi-agent systems (MAS) increasingly use latent collaboration to avoid the information loss and repeated encoding-decoding overhead of natu...
SUMONE is a lightweight summarizer designed to compress long interaction histories of language‑model agents. It is trained via masked reconstruction, where a frozen reconstructor predicts randomly masked parts of the original trajectory; reconstruction quality plus a length penalty yields a task‑agnostic reward for optimizing the summarizer with GRPO. At inference, only the summarizer remains, enabling plug‑and‑play context compression that reduces inference cost by roughly 78 % on DeepSearchQA and 75 % on SWE‑bench without affecting the underlying agent.
The paper introduces K-Token Merging, a latent-space compression method that merges each contiguous block of K token embeddings into a single embedding using a lightweight encoder. The compressed sequence is then processed by a LoRA-adapted large language model, while generation continues in the original vocabulary. Experiments on tasks such as structural reasoning, sentiment classification, and code editing demonstrate that K-Token Merging achieves up to 75% input length reduction with minimal performance loss, placing it on the Pareto frontier of performance versus compression.
arXiv:2606. 09659v1 Announce Type: cross Abstract: Long-context language model inference is bottlenecked by memory, as the KV cache grows with context length.