I really like Muse Wonder Full. Let me explain.

uses my lawn mow simulator in the youtube video... hits the bookmark button but doesn't even drop a like 😭😭 thats very cold.

not good at 3d. not good at frontend. not good at svgs and voxels either only good at coding which is not bad at all but why would someone pay for muse wonder full which cost way more then opus 5.5 (good at every fking single thing) just to code few things and get nothing else at all from model?

S - Opus 5.5,Fable 5.1

If muse wonder full really the better model, why didn’t you spell that full in this post

i cant believe mark, alex, and theo all agree on this

$50 plan after 40 minutes. Two agents. No subagents.

alexandr, mark, theo and i all agree on this :)

I wish I could try (at least in Muse the product) in the UK

thank you so much theo! excited to watch this rn!

S-tier on vibes is useful. The ranking I'd want next: $ per completed intent. Same models can flip once you score finished work, not screenshots.

the delegation part is very interesting, the subagent prompts are very good

It’s great! A bit slow, and I’ve been running into some compaction issues, but I love having a new contender in the space! Never thought I’d say that, but I’m genuinely rooting for Meta to succeed here!

Can’t wait to watch this one later. Very interested to see how it stacks up against the other models

It’s also less frustrating to speak with, less BS

The Full tier actually matching the demo hype is rare 🔥 Most "full" models look great in a clip and fall apart on a real refactor. Curious which part of Full sold you first: the long-context hold or the code quality?

Muse sub is so horrible usage limit. The model is not bad but not worth it. Better pay for Claude or Codex if you have limit budget. Don't waste money and valuable time for non-frontier.

legal, pena que ninguém vai usar. nós pobres não podemos nos dar ao luxo de gastar $50 ou $100 "só pra testar" um modelo, infelizmente. se eles querem migração em massa precisam oferecer um período free.

Muse is officially a frontier model But shit got really benchmarked

did the merge this instead behavior from your bug test show up again in the other tasks?

Okay explain how much you got paid.

Watched the whole video. Your reasoning is still pretty weak to put it in S tier.

We don’t even need to discuss whether you’re being paid, because that misses the real issue. The core problem is structural access bias. Early access isn’t charity. It is granted by AI labs that naturally protect their own commercial interests. Any creator who relies on that access faces an inherent incentive, conscious or not, to avoid burning bridges. If being completely candid risks losing future access, then editorial independence is compromised by design. People aren’t just being cynical. They recognize that protecting that professional relationship and ensuring future access inevitably softens the critique.

You use your own harness? My sol is not your sol

How is Muse Wonder Full inside their sub ?

Muse Wonder Full earning the S-tier explanation is the ranking that matters. Benchmarks are noise. "I actually like using this" is the real scoreboard.

Would love to try it, but its gated behind the 200 $ Plan 🥲

i need to know more about the model

Locking the best model behind a $200 sub instead of an API is the most Meta move possible, and now I kind of want it.

Wonder is new. Opus is the one people already trust when the code has to be right.

Stayed up longer to watch this. Was about to sleep until I saw it. Super exciting, and I hope they push even harder with a larger model, more post-training, etc!

You already had Muse Wonder in S. Did Full change how you use it day to day, or was this more about showing people why it's up there?