Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The landscape provides agent builders with a structured, citation-backed reference for selecting from 72 open-source memory systems, and highlights that MCP integrations already exist for most of them.
The paper provides a concrete, criteria-based framework for evaluating claims of recursive self-design in AI systems, grounding the discussion in publicly verifiable evidence from systems like DGM rather than treating MetaAI as an established paradigm.
FrontierCode's launch directly addresses the credibility gap in existing AI coding benchmarks — most notably the finding that over half of SWEBench results are unmergeable — by introducing maintainer-validated rubrics that measure real-world code quality rather than test-passing alone.
A leaked, unverified model called Oceanus V1-P outscored all other models tested — including Opus 4.8 and GPT-5.5 — by a wide margin on a diverse set of practical coding and reasoning tasks, though its true origin and stability remain unknown.
Omi Med STT v1 is the best-performing locally-running open model on this benchmark, achieving cloud-competitive M-WER at 0.6B parameters while keeping patient audio entirely on-device.
ALMANAC provides the first dataset with action-level mental model annotations grounded in authentic human collaboration, offering a concrete benchmark for evaluating whether LLM agents can simulate the reasoning alignment that effective human collaboration requires.
The work addresses the practical economic and computational constraint of LLM-call costs in counterfactual recourse, showing that a structured agentic search strategy can produce more diverse, validated alternatives without increasing budget expenditure.
The study reveals that the gap between stage-level and end-to-end pipeline automation in real scientific workflows is a distinct, underexplored challenge not captured by existing coding agent benchmarks.
The paper formalizes a conceptual framework — including the new discipline of "Agentic Engineering" and the AaaS category — that attempts to give researchers and practitioners a structured vocabulary for understanding how LLM-driven agents differ fundamentally from traditional software systems.
SWE-Explore provides a fine-grained diagnostic lens on coding agent capabilities that binary benchmarks like SWE-bench cannot offer, enabling targeted measurement of where exploration quality breaks down before the repair stage.