DeepSeek V4 architecture uses sparse attention to cut inference costs 73% at one-million-token contexts, but a NIST ...

Codex 这类命令行 AI 编程助手虽然强大，但直接用 OpenAI 官方 API 有3个痛点： 贵且慢：GPT-4 级别的模型成本不低，国内直连还经常超时； 换模型麻烦：想切国产模型，得手动改环境变量或配置文件，稍有不慎就报错。 协议限制：新版Codex只能接入Responses标准的API，而目前国内大模型基本都不支持。 作为vibe coding的重度用户，我几乎每天都在使用codex来完 ...

Researchers found 15 malicious JetBrains plugins posing as AI coding tools that exfiltrate OpenAI, DeepSeek, and SiliconFlow ...

At least 15 plug-ins for JetBrains IDEs transmit API keys to an external server, while otherwise offering their promised ...

DeepSeek has released the V3.2 and V3.2-Speciale models across web, app, and API. The company said V3.2 adds built-in reasoning for agent tasks and is its first model to support tool calls in both ...

After rocking the global AI and business community early this year with the January 20 initial release of its hit open source reasoning AI model R1, the Chinese startup DeepSeek — a spinoff of ...

BEIJING--Chinese artificial intelligence startup DeepSeek released on Thursday an upgrade to its flagship V3 model that the ...

A 75% reduction highlights falling inference costs and challenges premium pricing from OpenAI, Anthropic, and Google.

The reasoning model of DeepSeek goes through a chain of thoughts (CoT) to enhance the accuracy of its responses. The DeepSeek API provides users with access to the CoT content generated by ...

格隆汇4月25日｜DeepSeek官网更新API文档，页面宣布旗下新发布的旗舰级大模型DeepSeek-V4-Pro开启限时2.5折优惠活动。此次调价后，DeepSeek-V4-Pro的输入价格（缓存命中）低至0.25元（每百万tokens），输入价格（缓存未命中）为3元，输出价格为6元。此次优惠活动将持续 ...

DeepSeek funding round is now closed at $7.4 billion, but the deal structure reveals more than the money: China’s state AI ...

The latest DeepSeek model, V4, is expected to come out later this month -- potentially around Lunar New Year. And according to reports, it outperforms ChatGPT and Claude, particularly on tasks that ...