最新文章
ArXiv multiyear funding and the open-research infrastructure AI depends on
On September 23, ArXiv announced multiyear commitments to support it as an independent non…
AI meets ARM: why Linux on Snapdragon X2 and Mercury 2.5 both matter for on-device AI
Two developments this week point in the same direction: AI is moving closer to the device …
Claude discovers novel enzyme system: why AI-driven biology research is entering a new phase
Anthropic’s new life sciences group announced on September 23 that Claude discovered a nov…
Apple LensVLM shows how compressing long context into images can shrink vision-language costs
Apple’s LensVLM-9B introduces a different way to handle long-context vision-language tasks…
Mercury 2.5 benchmarked: 770 tokens per second but below-average intelligence on Artificial Analysis
Mercury 2.5, released in September 2026, is one of the most interesting speed-focused mode…
OpenAI GPT-6 Sol and Luna released: what the new flagship model changes for developers
OpenAI launched GPT-6, codenamed Sol and Luna, on September 23, 2026. It is the strongest …
Claude Opus 5.5 released: major performance gains, 40% lower cost, Anthropic frontier model sets alignment record
Anthropic officially launched Claude Opus 5.5 on September 22, 2026. It is the first model…
企业 AI 治理与合规实战：如何构建安全、可控且可审计的大模型应用体系
当大模型从“实验性工具”变成写入企业核心业务流程的生产系统时，AI 治理不再是法务部门的专利，而是技术架构的必选项。2026 年上半年，我们为三家金融企业和一家跨境电商搭建了完整的…