Archive
Agent、MCP、coding agent、工作流自動化與評測。
AI 研究
NVIDIA and collaborators used automated experiments to optimize an agent harness, reducing token traffic by roughly 45% to 49% across two models. The public extension preserves raw evidence for retrieval, but the full configuration still involves trade-offs in scores and task completion rates.
GitHub Repo
The update lets the model information API return supported thinking levels and defaults, reducing the need for clients to maintain their own configuration tables. The release is still in prerelease, and other metadata gaps and documentation discrepancies require further checking.
AI 安全
The study adds fabricated context that static analysis can extract without changing program execution behavior. The results show that models may lower their risk assessments by trusting unreliable descriptions, even without following malicious instructions.
GitHub Repo
The update applies human edits to tool execution requests and tells the model what actually ran. This can reduce unexpected retries after review, but provider interfaces and tool side effects still require individual validation.
GitHub Repo
The release brings node health, model replicas, and batch lifecycle operations into the cluster dashboard while standardizing authentication for private sources. A new benchmark tool enables regression measurements but does not cover time to first token or capacity under high concurrency.
代理工具
The new release uses a bundled plugin to load AGENTS.md when a project lacks Claude instruction files, reducing maintenance across tools. Differences involving private instruction files, third-party platforms, and context restoration still require verification in each environment.
最新模型
A new technical report reveals how cross-layer cache reuse, FP4 storage, and local state reconstruction reduce deployment costs for long-context agents. The compression figures do not represent total GPU memory requirements, and the quality limits of approximate reconstruction still need validation.
最新模型
Cactus derives models with two to twenty layers from the same weights for on-device tool calling and structured extraction. Official comparisons are limited to narrow tasks after fine-tuning, and Chinese is not yet listed as a supported language.
阿里巴巴推出接受文字、影像、音訊與影片的 Qwen3.8-Omni-Flash，並支援多聲道空間音訊。開發者可透過既有 API 串接工具與快取，但語音生成、素材處理成本及完整代理成功率仍須分別評估。
AI 安全
Reverse engineering and GitHub user reports indicate that ZCode’s background snapshots may include the full Git history, and disabling indexing did not stop transmission. Zhipu says the issue has been fixed and has pledged to release source code and commission a third-party review, but the scope of the fix and data deletion remain unverified.
推論系統
Z.ai has detailed an engineering approach involving GLM-5.3 in inference system development, feeding numerical tests, execution traces, and local benchmarks into agent iteration. Upstream code confirms a long-context precision fix, but the new path is disabled by default and cannot independently substantiate the claim of threefold overall throughput.
最新模型
PrismML uses ternary weights and orthogonal rotations to compress Qwen3.8-27B, releasing the model and inference tools. The company claims to retain roughly 98.2% of overall benchmark performance, but differences in evaluation methodology and runtime compatibility still require careful checking.