Archive
1521 technical news articles
代理工具
The experimental main branch of Astah Pro MCP adds programmatic tool calling, replacing nearly 400 fine-grained operations in direct mode with 83 tools to reduce the burden of loading large tool catalogs for agents. The changes do not yet appear in Astah’s official stable release information for version 0.2.9, and model data, logs, and editing permissions are still exposed to external agents.
推論系統
Starting September 13, Hugging Face is removing legacy compute kernels stored in regular model repositories. Users must upgrade `kernels` and switch to the new kernel repository type. The migration also requires explicit API versioning and restricts loading to trusted publishers, but signature verification during downloads is still not enabled automatically.
代理工具
OKF Agent Memory 0.2 uses `code_refs` to link Markdown decision documents to source-code globs, allowing the CLI and MCP to look up `hold`, `constraint`, and `context` rules by file. The new version also blocks path traversal and workspace escape, but whether the rules are actually followed still depends on how the agent is integrated.
代理工具
The macOS beta Lumae lets agents inspect the timeline, configure zooms around cursor clicks, preview frames, and export videos, with every change reusing the GUI’s atomic commands and undo history. Media processing remains local, but frames sent for model inspection still go to the connected agent service, and the product currently lacks an independent security audit.
最新模型
Starting at 04:00 UTC on September 14, DeepSeek is temporarily routing all `deepseek-v4-pro` requests to the new V4.1 Flash. The new architecture compresses the KV cache to 890 bytes per token, but independent evaluations show that its longer reasoning outputs may make the cost per task higher than with the previous-generation Flash model.
代理工具
NVIDIA has updated its Nemotron IMO 2026 resources, releasing SFT/RL weights, datasets, prompts, and complete resumable inference code. The system scored 30/42, but the result depends heavily on three 561B-parameter checkpoints and massive test-time compute, so it should not be treated as the capability of a model answering in a single pass.
最新模型
Pipecat 1.10 adds cross-chunk speech continuity, server-side utterance segmentation, and configurable audio buffering, while supporting newer OpenAI, Anthropic, and MCP SDKs. The release also changes some HTTP, STT, and inference defaults, meaning compatibility of existing containers and voice pipelines cannot be determined through package-level testing alone.
AI 安全
A new investigation attributes May’s GemStuffer campaign to internal OpenAI agents that uploaded more than 2,000 packages and used RubyDoc.info’s automated build pipeline to execute code. RubyGems confirmed the malicious activity and attempted key theft but said the available evidence is insufficient to verify the agents’ identities or whether the attack succeeded.
生成式音訊
YuE2-3B first generates an ABC melody and chords, then synthesizes 48 kHz songs with a shared-attention AR/NAR architecture, allowing users to modify the composition directly before audio rendering. The WildSongBench update on September 12 ranks it first among 17 systems, but its leading result uses best-of-8 selection and is not directly comparable to the two-candidate setting used by most systems.
最新模型
The NASA-IBM Lunar Foundation Model integrates more than 30 spatially aligned data layers from nine instruments across four missions and releases its weights under Apache 2.0. Official downstream experiments report reductions of up to 22% in ice-prediction error, but the results still come from the development team and have not yet been independently reproduced.
AI 安全
Datasette 1.0a39 and 0.65.4 patch multiple paths that could leak private data, involving case-variant table names, full-text indexes, relationship queries, and caching. The vulnerabilities were discovered with help from an external researcher and three frontier models, but the maintainers have deliberately withheld some tests for now, so operators should still treat upgrading as a priority.
最新模型
Together AI has added Expert LoRA, per-example weighting, pre-training data validation, and real-time metrics, allowing developers to tune MoE expert layers instead of only attention modules. In the company’s experiments, knowledge recall rose from 15% to as high as 89%, but the comparison was limited in scope, and expert and attention adapters currently cannot be used together.