LCM: Lossless Context Management

TL;DR

LCM introduces a deterministic architecture for long-context memory in LLMs that outperforms existing models on long-context tasks by using recursive context compression and task partitioning.

Contribution

The paper presents a novel, lossless, recursive context management system that enhances long-context handling in language models, extending the recursive paradigm with deterministic mechanisms.

Findings

LCM outperforms Claude Code on long-context benchmarks.

LCM achieves higher scores across context lengths from 32K to 1M tokens.

Recursive context manipulation can surpass frontier coding agents with file-system access.

Abstract

We introduce Lossless Context Management (LCM), a deterministic architecture for LLM memory that outperforms Claude Code on long-context tasks. When benchmarked using Opus 4.6, our LCM-augmented coding agent, Volt, achieves higher scores than Claude Code on the OOLONG long-context eval, including at every context length between 32K and 1M tokens. LCM may be considered both a vindication and extension of the recursive paradigm pioneered by Recursive Language Models (RLMs). Our results demonstrate that recursive context manipulation can outperform not just conventional LLMs, but frontier coding agents with native file-system access. LCM departs from RLM by decomposing symbolic recursion into two deterministic, engine-managed mechanisms: recursive context compression, in which a hierarchical summary DAG automatically compacts older messages while retaining lossless pointers to every original; and recursive task partitioning, in which engine-managed parallel primitives like LLM-Map replace model-written loops. This trade-off, analogous to the move from GOTO to structured control flow in program-ming language design, sacrifices maximal flexibility for termination guarantees, zero-cost continuity on short tasks, and lossless retrievability of all prior state.