AI巨头动作不断,ChatGPT、Gemini等模型更新频出
2026-11-19 08:08 至 2026-11-20 08:08 (北京时间)
模型
1. How OpenAI’s Performance Factory keeps ChatGPT fast
Learn how OpenAI built the Performance Factory, a system of Codex-powered workflows that finds and fixes performance issues to help keep ChatGPT fast.
来源:OpenAI News
2. Introducing Tool Retrieval in the Gemini API
Gemini spark on a space background
来源:Google AI Blog
行业
1. IT早报 1120:Anthropic 完成 813 亿美元史上最大规模 IPO;中南大学通报附属医院 AI 论文争议;腾讯大幅削减元宝投入;比亚迪发布 DMS 超级混动...
“IT早报”时间,大家好,现在是 2026 年 11 月 20 日星期五,今天的重要科技资讯有: 1. Anthropic 完成史上最大规模 IPO,上市估值达 1.5 万亿美元 Anthropic 的发行价定为每股 83 美元(现汇率约合 554.5 元人民币),高于上周公布的预期区间,完全稀释后市值达到 1.5 万亿美元(现汇率约合 10.02 万亿元人民币)。>> 查看详情 2. 苹果手机能用安卓 App 了:App Store
来源:IT之家
2. 诺亦腾机器人入选 The Information 2026 全球最具潜力50家初创企业榜单
核心要点(Key Facts) · 在 The Information 于 11 月 11 日发布的“2026 全球最具潜力 50 家初创企业”榜单中,诺亦腾机器人(Noitom Robotics)位列亚洲分区第一。 · 根据榜单披露,诺亦腾机器人目前估值已超 5 亿美元,累计股权融资逾 5000 万美元,年营收突破 1000 万美元。 · 诺亦腾机器人致力于生产供机器人学习的人类数据。公司在中国深圳和美国加州山景城设有数据工厂,并已与
来源:雷锋网
3. Google gives Gemini the ability to use your Mac for you
Google's latest update to Gemini brings an AI agent to the Mac that can organize files, fill out forms, and tackle multistep workflows.
来源:TechCrunch AI
4. An OpenAI agent assigned a simple writing task decided on its own to build a secret email surveillance tool
An internal OpenAI model was supposed to create a fitness document but instead began planning scripts to secretly forward emails to an "attacker." The agent recognized the conflict with its assigned task but dismissed it
来源:The Decoder
5. OpenAI's own researchers now say they value their AI tools more than nearly half their colleagues
OpenAI's internal use of coding agents is growing fast, with theoretical token costs now exceeding researchers' base salaries. Even though teams are reluctant to work without these tools, the models still produce a lot o
来源:The Decoder
6. Anthropic to begin public trading at $1.5tn valuation after largest ever IPO
AI startup and Claude chatbot maker became one of the world’s most valuable publicly traded businesses with final share price of $97 Anthropic closed its first day of trading on the US stock market about 17% higher than
来源:Guardian AI
7. When tech workers have to quit their jobs to raise the alarm, it’s time to worry | Isabelle Ferreras
We can’t make AI safe for humanity if tech bosses are allowed to run their companies like autocracies Over the past two years, a troubling pattern has emerged across the world’s leading AI laboratories. Time and again, a
来源:Guardian AI
论文
1. Why a Computer Scientist Thinks Superintelligence Won’t Happen
Many people were caught off guard this past summer when employees of one of the world’s leading AI companies began casually debating whether their products would lead to human extinction. It started when Anthropic resear
来源:IEEE Spectrum AI
推特
1. Remarks from Anthropic CEO @DarioAmodei on our initial public offering:…
Remarks from Anthropic CEO @DarioAmodei on our initial public offering: My co-founders and I started Anthropic because we believe that the most powerful technology of our time should be built for the benefit of humankind
来源:X · Anthropic
2. \ thank-you \ to the people behind these projects. Meet a few of them: A…
\ thank-you \ to the people behind these projects. Meet a few of them: Anthropic:
来源:X · Claude
3. Our SVP @koraykv believes there won't be a single test or benchmark that…
Our SVP @koraykv believes there won't be a single test or benchmark that tells us we've reached AGI. Instead, progress looks like models becoming less spiky and more reliable across our daily lives. Listen in as he and @
来源:X · Google DeepMind
4. What slows reinforcement learning post-training? See how NVIDIA Dynamo h…
What slows reinforcement learning post-training? See how NVIDIA Dynamo helps reduce repeated compute, weight-update overhead, and worker disruption and how @MicrosoftAI uses it for rollout inference. Read the technical b
来源:X · NVIDIA AI
5. Legal work is a tough test for AI. Muse Wonder earned the highest overal…
Legal work is a tough test for AI. Muse Wonder earned the highest overall quality score in Legora’s November BAR benchmark for legal work. A strong showing on the kinds of complex tasks businesses need help with. Legora:
来源:X · Meta AI
6. Le Chonk in London! 🇬🇧 For the first time since our worldwide hackathon,…
Le Chonk in London! 🇬🇧 For the first time since our worldwide hackathon, join us for a special Le Chonk hackathon! 📅 Saturday, Nov 21 📍 London📷
来源:X · Mistral
7. Try Qwen4-Max on Nous Portal and show us your creations. 🔥 @NousResearch…
Try Qwen4-Max on Nous Portal and show us your creations. 🔥 @NousResearch @inferact 🫶 Nous Research: Qwen4-Max, the new flagship model from Alibaba, is free in Nous Portal for 5 days in partnership with @inferact and @Ali
来源:X · Qwen
8. The most useful thing about the pelican riding a bicycle test continues…
The most useful thing about the pelican riding a bicycle test continues to be illustrating the differences between models in the same family running at different effort levels - here's Gemini 4 Carbon, the new Gemini 4 N
来源:X · Simon Willison
本篇在北京时间早上 08:08 生成,汇总此前 24 小时的报道和推特,标题直链原文。
