用户因 Muse 免费且额度宽裕退订 Claude

Jack Kubinec @whosknave原推发布于 2026.10.03

在 X 上打开原推

Muse is good enough that I've canceled my Claude subscription. I use ChatGPT for work tasks (mainly preliminary background research on topics I plan to write about), but I used to use Claude for personal projects like fitness tracking and becoming a better texter. Claude recently became very token-hungry; I was constantly hitting my usage limits by doing basic tasks on lower-effort models. Muse can seemingly do several multiples of the work Claude can do without hitting usage limits, and it's free! Muse is the first AI product that has felt immediately immediately useful for me as a non-coder. Other personal agents, such as Town, required too much setup and "botsitting" and seemed like they couldn't be helpful unless I logged my entire existence on Google Calendar. Muse just works. In the past day or two, Muse sold stuff that was sitting dormant on Facebook Marketplace (longtime followers know this is my favorite hobby), dealt with work reimbursements, canceled a bunch of subscriptions, and even found unclaimed money owed to me by my college. I don't have particularly rosy feelings about Meta, but I do think people underestimate it as a force in the AI race. There is a realistic scenario where AI's diffusion into the economy and society takes a lot longer than its proponents think, and OpenAI and Anthropic get too far over their skis on the data center buildout. Meta, which owns many of its data centers and is undeniably good at building products, could end up taking the AI consumer market and monetizing it through things like agentic commerce and serving ads on Muse. The NYT's reporting on Anthropic's courting of religious leaders was a good reminder that lots of people at frontier labs assign religious significance to the development of AI. They believe they are developing quasi-conscious beings that will soon exceed humans on all useful tasks and will either deliver a post-scarcity utopia or the apocalypse. Perhaps they are flattering themselves, and AI is less like a new being and more like a new tool. I think Zuck sees AI more in that light (although he too is prone to AI boosterism, as in the case of his weird recent blog anecdote about superintelligent lawyers). He's also someone who is ruthlessly focused on making good products—products that have nuked attention spans and made society worse, yes—but good products nonetheless. Muse is a good product. I wouldn't bet against Meta.