DeepSeek V4 发布时，在技术报告里写的非常真诚：在推理能力上，「落后前沿闭源模型大约 3 到 6 个月」。 而最近两天，我的各个编程交流群里就开始了对 V4 和各家模型进行大量比较、讨论。其中讨论到的国产模型，最多的一个我看下来就是智谱的 ...

点击上方“Deephub Imba”,关注公众号,好文章不错过 !Graphify 是一个 Python 工具，同时也是一个 Claude Code ...

DeepSeek V4 arrives in Pro and Flash variants with a 1M token context window, lower inference costs, and a stronger push into ...

OpenAI Releases GPT-5.5, a Fully Retrained Agentic Model That Scores 82.7% on Terminal-Bench 2.0 and 84.9% on GDPval ...

A new supply chain attack targeting the Node Package Manager (npm) ecosystem is stealing developer credentials and attempting to spread through packages published from compromised accounts.

当 AI 智能体不再只是「一次性工具」，而是能够持续学习、自我进化的「数字伙伴『数字同事』，会发生什么？自进化智能体应该采取怎样的设计原则？ 全球首个基于「上下文信息密度最大化」设计原则的自进化智能体系统 ——GenericAgent（GA），正式发布其技术报告。 报告显示，GA 能在保持任务准确率前提下，比同类竞争对手节省近 10 倍 Token。报告深度解读了 GA 的核心设计理念，介绍了自进 ...

A multi-tenant authentication gap in Microsoft’s AI operations agent exposed live command streams, internal reasoning, and ...

Claude has a workflow-breaking problem, and it's about time it is addressed ...

While Anthropic's dispute with the Pentagon escalated over guardrails on military use, OpenAI LLC struck its own publicized ...

Abstract: Large language models (LLMs) have shown exceptional performance in code generation and understanding tasks, yet their high computational costs hinder broader adoption. One important factor ...

Opus 4.7 utilizes an updated tokenizer that improves text processing efficiency, though it can increase the token count of ...