Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
HarnessX demonstrates that evolving the runtime scaffolding around a model — rather than scaling the model itself — can deliver substantial benchmark gains, offering a complementary path to agent improvement that does not require larger or more expensive models.
AgentSpec provides the first controlled compositional foundation for studying embodied LLM agents, revealing that scaffold interaction effects — not individual module quality — determine performance, which reframes how agent systems should be designed and compared.
The paper establishes a fundamental, mathematically proven ceiling on multi-agent system performance that is determined by task structure — specifically C_min — meaning agent scaling and increased communication cannot overcome poorly structured tasks.
The benchmark reveals that dialogue capability is a distinct dimension of coding agent performance not captured by existing autonomous-system evaluations, exposing a gap between how agents are benchmarked and how they are actually used.
GTBP directly addresses the two key failure modes of existing context adaptation methods — inaccurate credit assignment and lack of convergence guarantees — in multi-LLM agentic pipelines, providing both theoretical stability proofs and empirical gains across three benchmarks.
GLM-5.2's combination of a 1 million token context window, expected MIT-licensed open weights, and ~$8/month pricing places a near-frontier coding model within reach of developers who cannot afford or prefer not to use Claude or Codex pricing tiers.
DiffusionGemma's parallel token-generation architecture produces fluent but factually unreliable text, with error rates that grow as topics become more obscure — a concrete limitation that distinguishes it from its autoregressive counterpart for any fact-sensitive use case.
MiniPIC removes the requirement for identical prefixes to reuse KV cache entries, enabling efficient caching of recurring structured inputs in retrieval-augmented and agentic workloads without the large server-side code changes or host-to-device transfer overhead of prior PIC approaches.
In head-to-head agent workflow testing, Minimax M3 completed more tasks at roughly 5x lower cost than Kimi K2.6, directly challenging the assumption that higher-priced models deliver proportionally better results in production agentic systems.
For agentic workloads, the analysis shows that a model's per-token list price is a misleading cost signal — turn count and token volume at runtime determine the actual bill, making session-log auditing the only reliable way to compare model costs.