编辑
2025-05-19
Brief News
00

OpenAI发布Codex编程智能体,可并行处理多项开发任务

简报:

  • OpenAI于5月17日正式发布云端软件工程智能体Codex,现已集成至ChatGPT
  • Codex基于专为软件工程优化的codex-1模型,可并行处理编写功能、解答代码问题、修复bug、提交PR等任务
  • 每个任务在独立云沙盒环境中运行,预加载用户代码库,任务完成时间1-30分钟
  • 目前向ChatGPT Pro/Team/Enterprise用户开放,Plus/Edu用户即将支持
  • 基准测试显示codex-1在SWE-bench得分72.1%,优于Claude 3.7和o3-high

相关链接:

编辑
2025-05-18
Brief News
00

GitHub Copilot全面升级为智能体,微软推出自主SWE智能体

简报:

  • GitHub Copilot宣布全面接入智能体技术,推出Agent模式(预览版),具备自主迭代代码、识别错误并自动修复的能力
  • 微软首次公开自主SWE智能体(项目代号Padawan),可像人类工程师一样主动解决编程问题
  • 新功能Copilot Edits支持同时处理多个文件并提供代码修改建议,简化代码审查流程
  • 新增视觉功能,支持上传截图、照片或图表生成相应界面和代码
  • 微软CEO纳德拉表示此次升级将Copilot从"编程搭档"提升至"人类程序员水平"

相关链接:

GitHub Copilot Fully Upgraded to Intelligent Agent, Microsoft Launches Autonomous SWE Agent

Newsletter:

  • GitHub Copilot announces full integration of intelligent agent technology, launching Agent Mode (preview version), capable of autonomously iterating code, identifying errors, and auto-fixing them 🛠️
  • Microsoft unveils its first autonomous SWE Agent (project codename Padawan), which can proactively solve programming issues like a human engineer 💻
  • New feature Copilot Edits supports simultaneous handling of multiple files and offers code modification suggestions, streamlining the code review process
  • New visual feature allows users to upload screenshots, photos, or charts to generate corresponding interfaces and code 📸
  • Microsoft CEO Nadella states that this upgrade elevates Copilot from a "programming partner" to the "level of a human programmer"

Related Links:

Autonomous /ɔːˈtɑː.nə.məs/
adj. 自主的
"Microsoft introduces an autonomous SWE agent capable of solving programming issues independently."
[例句] 微软推出了一款能够独立解决编程问题的自主SWE智能体。
词根分析
auto-
自己
-nomous
法律,规则
衍生词
autonomy (n.) 自主权

编辑
2025-05-17
Brief News
00

MiniMax语音模型Speech-02登顶全球双榜,超越OpenAI和ElevenLabs

简报:

  • MiniMax最新发布的Speech-02语音模型同时拿下Artificial Analysis Speech Arena和Hugging Face TTS Arena两项全球权威语音基准测评第一
  • 该模型在字错率(WER)和说话人相似度(SIM)等关键指标上均取得SOTA结果,成为榜单前十名中唯一的国产玩家
  • Speech-02支持32种语言,最少只需10秒参考样本即可完成声音模仿,中文和粤语字错率仅为2.252%
  • 模型定价仅为ElevenLabs竞品模型的1/4至1/2,性价比称得上全球SOTA
  • 此前大火的"AI吴彦祖"和故宫AI向导等应用均基于MiniMax的语音技术

相关链接:

MiniMax Speech Model Speech-02 Tops Global Dual Rankings, Surpassing OpenAI and ElevenLabs

Newsletter:

  • MiniMax's newly released Speech-02 voice model has simultaneously clinched the top spot in two globally authoritative voice benchmarking evaluations: Artificial Analysis Speech Arena and Hugging Face TTS Arena. 🏆
  • The model achieved State-of-the-Art (SOTA) results in key metrics such as Word Error Rate (WER) and Speaker Similarity (SIM), making it the only domestic player in the top ten of the rankings.
  • Speech-02 supports 32 languages and can replicate voices with just a 10-second reference sample, boasting an impressively low WER of only 2.252% for Chinese and Cantonese. 🎤
  • Priced at just 1/4 to 1/2 of ElevenLabs' competing models, Speech-02 offers unparalleled cost-effectiveness, arguably the global SOTA in value.
  • Popular applications like "AI Daniel Wu" and the AI guide for the Forbidden City were previously built on MiniMax's voice technology. 🤖

Related Links:

Benchmark /ˈben(t)ʃˌmɑːrk/
n. 基准;标准
"MiniMax's Speech-02 model topped two global authoritative speech benchmark evaluations."
[例句] MiniMax的Speech-02模型在两项全球权威语音基准测评中名列前茅。
词根分析
bench-
长凳;基础
-mark
标记;标准
衍生词
benchmarking (n.) 基准测试;对标

编辑
2025-05-14
Brief News
00

A futuristic AI medical interface glowing with holographic health data, surrounded by floating diagnostic charts and molecular structures, in a sleek high-tech hospital setting with soft blue lighting.jpg

GPT-4o领跑AI模型市场,Gemini 2.5 Pro在推理领域崛起

简报:

  • Poe平台发布的2025年1-5月AI模型使用趋势报告显示,GPT-4o在文本生成领域占据35.8%的市场份额,保持领先地位。
  • OpenAI推出的GPT-4.1系列迅速获得9.4%的市场份额,而Google的Gemini 2.5 Pro在推出六周内占据了推理模型31%的份额。
  • 推理模型市场快速增长,使用比例从年初的2%上升至10%,用户更看重模型解决复杂问题的能力。
  • Anthropic的Claude模型份额下降10%,DeepSeek R1的份额从7%降至3%。
  • 图像生成领域竞争激烈,GPT-4o的多模态生图功能引发广泛关注,Midjourney即将推出V7版本应对挑战。
Inference /ˈɪn.fər.əns/
n. 推理;推断
"Google's Gemini 2.5 Pro has taken 31% of the inference model market within six weeks."
[例句] 谷歌的Gemini 2.5 Pro在六周内占据了推理模型市场31%的份额。
词根分析
in-
向内
-fer-
带来
衍生词
infer (v.) 推断;推论

编辑
2025-05-09
Brief News
00

20250509110050653abe5f94b843e1_0.jpg

OpenAI任命Fidji Simo为应用CEO

简报:

  • OpenAI宣布任命Fidji Simo为应用CEO,负责领导产品、业务和其他公司职能,直接向CEO Sam Altman汇报。
  • Altman将继续担任OpenAI CEO,专注于研究、算力和安全等核心技术领域。
  • Simo目前是Instacart CEO,将在未来几个月完成交接后正式加入OpenAI。
  • Simo曾在Meta担任Facebook应用负责人,并在Instacart成功带领公司上市并扩展广告业务。
  • OpenAI预计到2029年,免费用户变现和广告等业务将带来250亿美元收入。