AI is more than just chatting: Gemini takes over tasks, GPT does interaction, and Claude changes animations | Beyond Chat: AI at Work

Original author：大雷早上好

Original source：https://www.youtube.com/embed/bYrdISLj87Q

What interests me most this time is whether I can complete the next step smoothly after getting the AI ​​answer. From ChatGPT's interactive bicycle to the task division of Gemini work agents, let's see how Claude turns data into dashboards and modifiable animations. In this issue, along with the public demonstration, I will talk about what I will try first and what areas I still have to take over myself. Ultrafast's "up to 8 times" refers to token generation, compared to Sol Standard; not the entire job is 8 times faster. API short contexts are priced at US$12 per million input tokens and US$60 per million output tokens. Long contexts are priced separately. The qualifications for API and Codex/ChatGPT Work should be looked at separately. Argon is still open in phases, and Claude’s new features are also beta and account-wide. Which step do you most want to hand over to AI first? 00:00 After AI answers, what can be done next? 00:29 Argon announcement and open scope 01:02 Gemini working agent and model selection 02:33 Ultrafast: How to calculate speed and cost 04:05 ChatGPT interactive answer 05:13 Claude Dashboards query and billboard 06:20 Claude Motion: The animation can continue to be modified 07:46 Try a familiar small task first 08:21 See you in the next issue 📰 Free AI Practical letter: https://dailycosmos.net/newsletter ☕ Ask Dalei for a cup of coffee: https://buymeacoffee.com/dalei ⭐ Join the channel membership: https://www.youtube.com/channel/UCk9tu0mFtXj_rOEfIncxuJQ/join 📚 AI solution library: https://aiagentclub.notion.site/ 🛠️ My works and practical projects: https://github.com/paul010/dalei-youtube Music: Elektronomia - Sky High [NCS Release] https://youtu.be/TW9d8vYrVFQ English summary A Chinese AI news roundup on Gemini work agents, interactive ChatGPT answers, GPT-6.1 Sol Ultrafast, and editable Claude dashboards and animations. Public demonstrations and my interpretation; not my own performance benchmarks. The up-to-8x claim token generation versus Sol Standard, not an entire task.