Context Language Models manage their own context as a file, beating SOTA with 21.5% fewer FLOPs

Researchers introduce Context Language Models (CLMs), which treat context as a file the model can update freely, learning what to keep. Built zero-shot with existing models, CLMs outperform state-of-the-art context management: 11.4% higher accuracy with 21.5% fewer FLOPs on BrowseComp-Plus, 5% higher scores with 59% fewer FLOPs on 12-hour EdgeBench, and 65% greater improvement on a 24-hour multi-repository agent-swarm task. Natural-language instructions and online RL further boost performance, while Suffix Cache Reuse cuts serving compute by 35%.

We introduce Context Language Models (CLMs), language models that natively manage their own context.
  1. _jayhack_

    Letting a model manage its own context is very bitter lesson-pilled

    Biggest challenge is you will get a much lower cache hit rate if you frequently edit the agent's context/prefix, so this can not be implemented efficiently via e.g. the Anthropic API.

    This ^ can be solved in principle but likely requires modifications to the transformer architecture and definitely to serving infrastructure

    See related: "KV Cache Rules Everything Around Me": https://www.completeskeptic.com/p/kv-cache-rules-everything-...

  2. bob1029

    I would be concerned with context management consuming limited attention resources.

    Do you want your agent solving its own memory crisis, or do you want it solving the actual task? It can probably do both at the same time, but I suspect there is a non trivial cost associated with this.

    A separate hypervisor agent that manages the main agent's context would be much better in my experience. You can run it on a different schedule and the main agent has to spend zero tokens thinking about it. This also makes it a lot easier to control when caches will be missed.

  3. killerstorm

    Related: "Recursive Language Models" https://arxiv.org/abs/2512.24601

  4. svachalek

    Wow. Context management is one of the big remaining hassles with modern LLMs so this could be big. The obvious complication is cache busting so it's also exciting they investigated solutions for that.

  5. Bolwin

    The biggest discovery might actually be that they ignored regular caching rules and kept invalid cache suffixes and it didn't hurt performance

More from this day

2026-10-01