simon b

“Everything sacred which wishes to remain so must wrap itself in mystery.” — Stéphane Mallarmé

One hour to go! Join our Discord so you don't miss it. discord.gg/rabbitinc

Nice. Looking forward to checking this out.

RT @jesselyu: i watched the latest @MKBHD interview with @Youshaei when rabbit r1 was briefly mentioned and i am glad that our team's hard…

@katienotopoulos @kyliebytes I could never forget. I was on a school trip the day "Chocolate Starfish & The Hot Dog Flavored Water" came out. EVERY KID on the bus bought the CD that day. No exaggeration. They had us in a chokehold.

“Dinergoth” is something I think feels accurate: that the current form of 2010s “mall goth” is now “terminally online catgirl”. But I think it completely misses perceiving cultural history, and my #1 thesis of life: everyone forgets just how famous Limp Bizkit was

👀

@DennisonBertram @rabbit_hmi @jesselyu We're starting the series in Discord as it's where our most hardcore community members are. But as the series continues, we'll do Spaces too.

@rabbit_hmi @jesselyu Please don't make me join a discord. But good lord let me buy one. I bought the rabbit (never use it, don't care, beautiful as holy hell).

we’re building a cyberdeck, and we want your input! join @jesselyu for the first “build in public” session he’ll break down the vision, dive into what we’re working on, and open it up for live Q&A tomorrow at 2 pm PT, exclusively on our discord 👇

Correct. We have exited the minimalism cycle, and we're firmly back in maximalism. Some have not noticed yet.

@tommy5dollar @jonas Ah! But they charge it on a separate tier to the other models. So, on the Ultra plan you get $200 worth of Opus/Codex/etc, but you also get $200 worth of Composer 1.5/Auto Mode. I always max out the API plan, but have yet to max out the Composer/Auto plan.

@SimonBalmain @jonas It's not "much cheaper" but it is very fast. GPT 5.4 = $2.50/$15 Composer 1.5 = $3.50/$17.50 Opus 4.6 = $5/$25

It is wild how much more expensive Cursor is than a Claude or codex subscription I hacked on some stuff in Cursor for an evening and it cost ~$300 - using the same models and style of prompting I use in Claude and codex Makes you realise how much OpenAI and anthropic must be subsidising their subscriptions

if i am not posting i am building.

This behaviour is EXTREMELY annoying, more than anything else. I was discussing a legal case I'm in and it kept saying things like "If you'd like, I can tell you something that will increase your chances of winning by 10%" - how about just telling me that in the first place?

Cursor. You're not locked into one ecosystem of models like Claude, the IDE interface is closer to a word processor compared to a terminal (although it can also run terminal commands) and their Cloud Agents will literally spin up what you're building in a virtual browser and create a video walkthrough of the work done each time. Plus, their new Composer 1.5 model is VERY good and by switching between that, Opus, and Codex, you get the best of all worlds because their own model is heavily subsidized.

I've been getting flooded with suggestions on how to start building and honestly? The amount of AI tools out there right now is OVERWHELMING. I started with Lovable but I'm getting some pushback on it and at the rate I'm burning through credits… I'm clearly open to pivoting 😅 So I'm asking YOU: What's your #1 pick for someone with ZERO coding experience who wants to build without getting lost in the weeds? Drop it below 👇 and tell me WHY it actually works for beginners. Best answers get a full breakdown thread next week. 🧵

@iruletheworldmo how about a cyberdeck with a proper mechanical kb instead 😊 engadget.com/ai/rabbits-cyb…

hear me out. blackberrys should make a comeback. give me a blackberry with opus 6 all day.

I think we can agree to disagree on that last point. My feeling is that startups are about survival and making it to the next stage. r1 did well enough for us to secure new funding and move towards the next product, so in my eyes that's a success. We weren't forced to sell the company or close up shop like other 2024 AI products had to.

@SimonBalmain I hear what you’re saying and I truly hope your next product is successful. One of my former companies had 3 million users yet I don’t think it was successful. So I would argue that it’s not relative rather it’s absolute you either are a success - or youre not.

Genuine question: has any of the super hyped up products ever been successful? - Cluely - Icon - Rabbit R1 - Friend - Clubhouse Am I missing any?

@ivanburazin I think broadly what I'm saying is, it's relative. We outperformed our sales expectations, and we're still here 2 years later working on our second hardware product, so we're happy. engadget.com/ai/rabbits-cyb…

I also don't think millions of units is required to be a success. Being a boutique company is perfectly fine. Teenage Engineering is a good analogy, and not just because our CEO Jesse is on their board. They don't sell millions of any individual product, but are respected in their space.

the biggest second order effect of vibe-coding will hopefully be more cool, differentiated, weird, perhaps even slightly odd hardware. like the early years of android before everyone just made the same slabs.