Video yükleniyor...
Video Yüklenemedi
Figure is planning to release its own proactive AI assistant this week! Here is what we know so far 👀 > Hark will likely run on Opus 5.5 for general tasks, with Gemini 3.1 Flash Image, Qwen3.6-27B, and GPT-6 Astra referenced in the code for other specific tasks. >... show more
24,383 görüntüleme • 1 gün önce •via X (Twitter)
16 Yorum

>"Hark has the power to plan 800 weddings, buy 1000 grand pianos, move across the country 1,500 times, and hire 1,000 lawyers for when it all goes wrong." Yeah, I really need this. 😜

😅 I hope it will go beyond booking a hotel

Yes, 10,000 holidays. 😀

Documented 🗞️

pro four is a funny plan name

"memory and soul" is a bold combo for an agent whose first job is to earn your trust 🩺

One Memory and Soul sitting under Opus, Gemini, Qwen and GPT is a shared write surface. A later model will act on a preference an earlier model wrote, with no log that names the author.

Lately I’ve been missing a lot of updates fr 🥲

opus, gemini, gpt-6 astra, and alibaba's qwen in one assistant's code. the american AI stack is a group project now and qwen is the kid who actually did the work

That plan tier list feels a bit overwhelming already

the 20x usage plans lose me. i use grok bot most days and still couldnt tell you what id burn 20x on. paying for one chore it actually finishes every time makes way more sense to me

Curious whether Hark's Soul keeps one voice when a task hops from Opus to Qwen to GPT-6 Astra. Routing across that many models is where an assistant's personality usually gets inconsistent.

Ok but I hope they don't forget about Figure robots🤖

Easy to forget Hark isn't a side project: a $700M+ Series A in May and the Handoff browser agent in August. With Opus 5.5 underneath, the bet is the memory and proactive layer, not the model. #AI #AIagents

Of course no Android

Figure is racing Muse with a waitlist, free paid seats for the first 100K, and computer use built in. The robot company just entered the assistant wars.


