self-improvement
The day's most-upvoted paper turns a model's own routing logs into its next training set News
A team calling itself NeoHorse released NeoHorse-1, a system that records which model handled each user request and how it went, then converts those logs into training data for the next round, closing an evaluation-to-training loop the authors present as a prototype path toward recursive self-improvement.
AI agents built 18 versions of their own infrastructure and not one ever saved its work News
A benchmark called HarnessDev had six frontier models build and improve their own agent harnesses, and found that while all 18 code harnesses implemented an execution loop, only one checkpointed periodically -- and across 26,679 recorded trajectories, not a single checkpoint event occurred.
A frozen model can look like it taught itself, and most self-improvement results never checked News
A new audit ran a completely untrained control model through the same self-training pipeline as the real thing and found it appeared to both learn and forget, meaning most reported self-improvement gains are measurement artifacts unless the null was measured too.
A robot system that improves without touching the model News
Researchers at Tsinghua froze the robot's underlying policy entirely and improved performance by evolving the scaffolding around it instead, shipping versioned packages of critics, recovery playbooks and tools rather than new weights.
EXO keeps an agent's memory outside the code the agent rewrites News
The EXO agent runtime splits a self-modifying agent into a disposable policy layer and a durable state layer, so an agent can rewrite its own prompts, tools and executor code without being able to damage its own event log, secrets or history.
A 9B model writes agent upgrades as good as Claude Opus 4.6 News
A 17-author study separates the ability to improve an AI agent's scaffolding from the ability to benefit from the improvement, and finds that a 9-billion-parameter model produces upgrades yielding gains comparable to Claude Opus 4.6.
SkillZip compresses an agent's skill file without ever running the agent News
A new method compresses the accumulated skill files of self-evolving agents by finding the shortest structural explanation that preserves every contract element, with no test rollouts required.
A model improved itself by training only where it disagreed with itself News
U-OPSD drops the teacher entirely: it samples several attempts, takes a majority vote as a stand-in answer, then trains only on the attempts that disagreed with that vote, matching or beating methods that use real ground-truth labels.
Ouroboros Tool
An agent harness that improves its own tools, prompts and core implementation through reviewed commits, which then become the runtime for its next task. Public code, with benchmark campaigns run on frozen snapshots so the numbers mean something.
NeoHorse-1 Tool
Code and models for a routing harness that logs which model handled each request and how it went, then converts those logs into the next round of training data. Released alongside the paper as a working implementation of the evaluation-to-training loop.