former trader retweeted
Meta just dropped Muse Ember 🔥
The smallest Muse model ever — and it was trained by Muse Wonder, the same model that runs my assistant.
It matches Claude Haiku's benchmark score at 1/3 of the cost. $0.10 per million tokens. 1M context window.
Wonder plans, Ember does the volume. That's how you put an AI agent in billions of pockets.
Day 5 of running my whole life on Muse — and they keep shipping.
former trader retweeted
Babe, your model is the perfect size. The big ones don't fit (in my VRAM).
Go check out muse ember. It might be small, but it knows how to satisfy you.
former trader retweeted
Replying to @bruno_munhoz @Muse
I just subscribed to the $20 tier yesterday. I've been using Muse since day one and the experience has been fantastic. Yesterday, I suddenly wanted to try it out with a better model.
I'm increasingly feeling like Muse is going to capture the personal AI agent market. Its email and calling features, along with the Space functionality, are all working brilliantly together.
I'm really looking forward to the new features they'll be rolling out next.
former trader retweeted
3 full days using @muse wonder and I hit 20% of limit
I have a lot of routines going on, with shorter intervals than I used to have on grokbot
At this pace I won’t even hit 50% of limit 😂
I LOVE IT
former trader retweeted
people are sleeping on @Muse Spaces
think of them like dashboards or personal apps, here are ones i've found useful so far:
+ family HUD= weekly schedules for everyone, can tag which parent is handling rides
+ personal health tracker= i log my weight daily, have a goal, and Muse pulls workouts from my connectors
+ content studio= ideas for future videos or write ups based on my interests, what's trending, and the performance of my IG posts
+ content queue= clips and posts waiting for me to schedule or edit. *Bonus on this one if you want to post to Instagram, since Muse has a great native connector
I set this up this weekend. It’s…just. So. Damn. Good.
Replying to @alexcornell
Can Muse adapt it based on factors like location, time of day, etc? For example show me different apps during the week vs weekend
former trader retweeted
You can design the entire Muse app just for you. With last week’s launch, you can make a space for anything you care about. And you can rearrange the app’s core navigation to create your perfect app.
Personal software.
I’ve got mine set up to support different personal gears: I have the app orientated around family and golf on the weekends, and work and news during the week. You don’t need to do anything fancy to get this working – just ask your Muse to set things up how you like. You can even ask for the tab bar to be removed, if you prefer.
In this video, you can see one of my spaces which I set up around memories. I wanted a place to reminisce and just asked for a daily grid of my Instagram stories. Simple, but it makes me smile every day.
I remember when we first prototyped this idea of a malleable app back in March. It was one of the most excited I’ve ever been about product development and it remains so.
former trader retweeted
pareto frontier who? 🏎️ 💨
- muse ember 🔥
BREAKING: Muse Ember by @AIatMeta has redefined the Preference vs Speed Pareto frontier on Overall HTML, knocking out 10 top models from @GoogleDeepMind, @OpenAI, and @DeepSeek_ai.
At 33s per generation, Muse Ember is 9.5× faster on average than the previous Pareto frontier models. It is also at least 34× cheaper on average than all the previous Pareto frontier models it replaced at $0.10/$0.50 per 1M input/output tokens ($0.004 cache). It is also ranked 10th overall with an Elo of 1345!
A huge congrats to the @AIatMeta team on a top-10 placement on the leaderboard!
former trader retweeted
This is how Meta scales Muse to more free users. Remember Zuckerberg saying they would improve cost-efficiency by 10X within a year?
If you’re bullish $META, LIKE this tweet for a big surprise…
former trader retweeted
Wait till you see what we been up at @Meta :)
Across our projects @GoogleDeepMind, we are witnessing a genuine phase transition in the pace of scientific discovery. You will see a glimpse of this in our releases in the next few weeks.
I believe the TLDR here is: $META is MASSIVELY undervalued.
I had been thinking about $META vs $ANT which have taken very different routes to the same valuation. But their respective set up appears quite different.
full piece (no paywall): mbi-deepdives.com/ant-vs-met…
former trader retweeted
Think about it like, if Meta can squeeze that much intelligence per unit of compute for a small model, it’s reasonable to assume they can accomplish the same for a larger model.
former trader retweeted
Oh wow. 👀
Muse Ember is the new world’s best smol & cheap model.
Move over Luna and Haiku!
Makes me bullish on the larger models Meta and Zuck have in the oven.
Releasing Muse Ember today -- our smallest model that delivers higher intelligence per dollar than others in its class. Built for token-heavy work that agents repeat all day: tool calls, coding subtasks, routing, extraction, and long document parsing. Available today on the Meta Model API and in Muse Code.
The Muse Ember model is truly elite from a cost-to-performance perspective. $META is cooking like you’ve never seen before.
They are coming for THE WHOLE market.
- consumer with @Muse
- enterprise with Ember and many others
Releasing Muse Ember today -- our smallest model that delivers higher intelligence per dollar than others in its class. Built for token-heavy work that agents repeat all day: tool calls, coding subtasks, routing, extraction, and long document parsing. Available today on the Meta Model API and in Muse Code.
former trader retweeted
Okay, I’ve now spent the last 24 hours tinkering with Muse Spaces.
I have 3 live “apps” going.
One tracks my fitness and calories.
One tracks my finances and portfolios in real time.
One hunts shopping deals across a few sites and ranks them by discount. I call it… “Only Deals”
All 3 built by just describing what I wanted.
No code. No review. No waiting weeks.
They are extremely simple to set up, pretty reliable, and super fun to play around with.
Think about that for a second.
Anyone can now build their own “app” in minutes, for exactly how they live and work.
That is not a feature. That is a platform shift.
Spaces are an App Store level opportunity for Meta.
Muse is the new operating system.
Zuckerberg might have finally figured out how to get out from under Apple.
Once he builds a phone, of course.
Been playing with Muse Spaces for the last several hours and, wow!
Meta essentially launched customizable mini apps.
It’s a big idea, not necessarily a new idea.
But Meta giving this capability away for free is what really stands outs.
Meta is drastically undercutting the other players, giving out token amounts and features previously reserved for Pro+ level plans.
Right now that undercutting is gated to only consumer use of AI.
My guess: this becomes a much bigger enterprise AI narrative in 2027.
And Meta is not alone.
There are a lot of well capitalized companies ready to undercut the current leaders. Clearly they still need to catch up on a performance basis, but if you assume a few get there, it is going to be a very interesting period for AI.

































