AI 战略简报 · 2026-09-24

↓ 下拉刷新

🔥 Gemini 3.5 Flash 发布

2 条最高 9.0 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2026-05-20 01:54

9.0

Google I/O 现场直击：我们正式推出 Gemini 3.5 Flash，这是我们在智能体（Agents）和编程任务上具备前沿性能的最新模型，目前正在全球范围逐步推出……

诺姆·沙泽尔@NoamShazeer ✓ · 2026-05-20 01:54

8.4

回复：3.5 Flash 的各项指标十分显著：在每秒输出 token 速度上是其他前沿模型的 4 倍，并在几乎所有评测中超越了我们先前的 3.1 Pro 模型……

🔥 Gemini 2.5 发布

1 条最高 7.8 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2025-03-26 01:11

7.8

转发：该模型的“思考”（thinking）能力带来了重大提升：在数学和科学基准测试（如 AIME、GPQA）中取得顶尖表现，在代码能力方面表现出众……

🔥 Gemini 2.0 Flash Thinking 更新

1 条最高 7.8 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2025-01-22 07:26

7.8

回复：我们在具有挑战性的数学、科学和多模态推理基准测试中提升了性能（AIME：73.3%，GPQA：74.2%，MMMU：75.4%）。我们还扩展了……

🔥 Gemini 2.5 Pro 发布

1 条最高 7.2 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2025-03-26 01:11

7.2

回复/转发：在此示例中，Gemini 2.5 Pro 编写代码生成了这个动态气泡图，展示了各国经济与健康指标的演变历程……

🔥 Gemini 2.0 Flash Thinking 发布

1 条最高 6.8 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2025-01-22 07:26

6.8

回复：请继续提供反馈！立即通过 https://goo.gle/4jsCqZC 体验新版本。

🔥 Gemini 新版本发布

1 条最高 6.4 @NoamShazeer

诺姆·沙泽尔@NoamShazeer ✓ · 2025-03-26 01:11

6.4

回复：我们非常期待看到大家用我们最新发布的 Gemini 模型构建出什么成果。我们将根据大家的反馈持续迭代！今天就前往 AI Studio 免费体验吧……

📘 AI Agent 架构

1 条最高 7.4

诺姆·沙泽尔@NoamShazeer ✓ · 2026-05-20 01:54

7.4

回复：性能与速度的结合使该模型成为长流程和 Agent 任务的理想选择。我们的企业合作伙伴已经在实际场景中看到切实成果……

📘 大模型训练技巧

1 条最高 7.6

诺姆·沙泽尔@NoamShazeer ✓ · 2025-01-22 07:26

7.6

回复：我们的目标是让模型更加可靠和一致，同时继续减少最终回答与思考过程之间的矛盾。

尚未被主编选入事件/主题的原始推文（共 20 条，按时间倒序）

诺姆·沙泽尔 @NoamShazeer08-20 01:40

RT Dean W. Ball: By the way if you are a young writer and not routinely freaking out about “whether you still got it” you aren’t doing it right. Th...

诺姆·沙泽尔 @NoamShazeer06-18 08:15

I’m excited to share that I’ll be joining OpenAI and look forward to working with the exceptional team there. It was a difficult decision to move on...

诺姆·沙泽尔 @NoamShazeer03-26 23:50

Gemini 3.1 Flash Live is now available, and it’s built for production-ready reliability. We’ve improved its overall quality so developers and enterprises can build voice-first agents that complete complex tasks at scale.
It leads on ComplexFuncBench Audio for multi-step function calling under constraints, and on Scale AI’s AudioMultiChallenge for long-horizon reasoning (with interruptions!). It's also inherently multilingual, powering the rollout in Search Live.
If you're building agents that need to execute complex tasks at scale, give it a spin.
https://t.co/Gfzo5S3hkk

诺姆·沙泽尔 @NoamShazeer03-04 03:03

📢Introducing Gemini 3.1 Flash-Lite, our fastest and most efficient model, built for high-volume workloads. It outperforms 2.5 Flash in reasoning, reliability, and scalability at a lower cost.

This model also introduces thinking levels. You can adjust compute by complexity of the task, burning zero thinking overhead on high-volume tasks, while reasoning through the complex edge cases.

Maximum intelligence, minimal latency.

Read more: https://t.co/FRhfbzigbv

诺姆·沙泽尔 @NoamShazeer02-20 00:22

Last week we upgraded Gemini 3 Deep Think. Today, we’re shipping the core intelligence that makes those breakthroughs possible: Gemini 3.1 Pro.
A noticeably smarter, more capable baseline for your hardest challenges.
Available now: https://t.co/Srm2Ec7OWJ https://t.co/GbOjtknyDL

诺姆·沙泽尔 @NoamShazeer02-13 00:43

An updated Gemini 3 Deep Think is out today:
📈 Achieves SOTA on ARC-AGI-2, MMMU-Pro, and HLE.
🥇Gold-medal level on Physics & Chemistry Olympiads.
It turns out the best way to solve hard problems is still to think about them. Read more: https://t.co/PMmuyRq90B https://t.co/cxCAbcEz7q

诺姆·沙泽尔 @NoamShazeer12-18 00:16

Gemini 3 Flash is live. ⚡️
We’ve packed Gemini 3’s Pro-grade reasoning into a leaner model with Flash-level latency, efficiency, and cost.
It's my favorite model to use – the latency feels like a real conversation, with the deep intelligence intact.
Available in the API, Gemini App, and Search. Give it a spin.
https://t.co/FeGrVUhLvN

诺姆·沙泽尔 @NoamShazeer12-05 04:34

Gemini 3 Deep Think mode is live for Ultra users today. When using parallel lines of thought in this mode, Gemini shows meaningful improvement on key reasoning benchmarks such as ARC-AGI-2 & HLE. I think you will be deeply impressed.
https://t.co/5uyvH43yzf

诺姆·沙泽尔 @NoamShazeer11-19 15:47

Huge congrats to the team on an amazing model. Thousands of contributors. Billions of use cases. Feeling the AGI! https://t.co/Adpolq33aX

诺姆·沙泽尔 @NoamShazeer05-14 07:58

Amazing feature, makes it so easy to collaborate and build. Using 'Copy Canvas' to add hand gestures...
https://t.co/orvxTtOQ36 https://t.co/GTzw9QHfkT

诺姆·沙泽尔 @NoamShazeer03-26 01:11

Introducing Gemini 2.5 Pro Experimental.

The 2.5 series marks a significant evolution: Gemini models are now fundamentally thinking models.

This means the model reasons before responding, to maximize accuracy -- and it’s our best Gemini model yet.

Blog - https://t.co/Gx35wlkomq

诺姆·沙泽尔 @NoamShazeer01-22 07:26

Your feedback on Gemini 2.0 Flash Thinking has been incredible—thank you!

We’ve taken your suggestions and made an experimental update…

诺姆·沙泽尔 @NoamShazeer12-20 01:00

We’ve been *thinking* about how to improve model reasoning and explainability

Introducing Gemini 2.0 Flash Thinking, an experimental model trained to think out loud, leading to stronger reasoning performance.

Excited to get this first model into the hands of developers to try out!

诺姆·沙泽尔 @NoamShazeer08-02 06:40

RT @character_ai: Thrilled to share that we're open sourcing our innovative approach to prompt design! Discover how Prompt Poet is revoluti…

诺姆·沙泽尔 @NoamShazeer07-05 14:13

🇺🇸 Happy 248 America! 🇺🇸

This is one of my favorite lines ever:

"We hold these truths to be self-evident, that all men are created equal, that they are endowed by their Creator with certain unalienable Rights, that among these are Life, Liberty and the pursuit of Happiness."

And this is how I would have written it:

I believe that God created every human with infinite, unique and incomprehensible value. As we create a powerful new entity, it is essential that the new entity hold sacred the life, liberty and agency of every individual human.

诺姆·沙泽尔 @NoamShazeer06-28 01:13

Incredibly proud of the team for our official launch of Character Calls!

诺姆·沙泽尔 @NoamShazeer06-20 22:03

Character AI is serving 20,000 QPS. Here are the technologies we use to serve hyper-efficiently. [https://t.co/R14Jt9Z5yo ]

诺姆·沙泽尔 @NoamShazeer02-28 14:56

https://t.co/abQu8pN4WJ
Check out my first blog post.

诺姆·沙泽尔 @NoamShazeer11-21 03:25

As a friendly competitor in the AI space, we share a core mission of building AI technology that benefits every person. While we don’t know the specifics, we are saddened to hear about the recent events at OpenAI. We send our best wishes to everyone involved.

诺姆·沙泽尔 @NoamShazeer11-04 12:23

https://t.co/LDnsI92UPw #characterai