We launched 🍉 (Muse Wonder) and 🍉 Ultra (Muse Wonder Full) today! What excites me most is how much closer we’re getting to agents you can actually delegate work to. Personal agents that understand you and your life, and professional agents that can take on complex, long-running projects. 🍉 is designed for personal and professional agent use Pretty Wonder Full to see this come together just a year after starting the agent team at TBD Labs. Huge credit to the incredible team that made it happen!

Yes we just don’t like markdown table or list and don’t like the model to ask a question back

“Muse 强就强在浏览器操作上。它是在真实浏览器里一步步点，登录、验证这种关口过不去就叫我，我上去登一下，登完剩下的全是它的活。” cheers! 🍻

muse wonder full on webdev arena not bad, but clearly lots of room for us to improve!

iPhoneDuoBench is totally next level 🏎️💨

？？？我怎么没想到

Excited to see Muse Spark Open released! 🚀 Proud that vLLM powered RL inference during its training at Meta. A glimpse of recursive self-improvement: RL makes models better; LLMs help us optimize inference; faster inference enables more RL. The flywheel keeps spinning. 🔁

Muse Spark Open -- weights releasing today -- runs great on new Mac Studios thanks to work with @johnternus and Apple.

muse spark open -- weights out today! best US open-source model, and runs on the new mac studio 🍎

Muse Spark Open was literally as good as the best model we released just two months ago. Now it’s open-weight. Enjoy! research.meta.ai/blog/muse-s…

In our health workstream at Meta we did a project where we wanted to see what research papers medical experts could write using a Muse Wonder coding scaffold that could access a database of 5M electronic health records we licensed from a vendor. The goal is that with 5M patients in this database, AI could do research to answer arbitrary questions like “did people who take statin have higher rates of diabetes later in life”? The thing we were betting on here is on what happens when you put human experts, a strong coding model, and access to data all together We worked with a small group of medical experts to help them use this AI tool and the punchline is that they all found that this setup enabled a much faster pace of health research. They generated ten manuscripts in around two weeks, with one researcher (Paddy Ssentongo) generating three. Throughout the process, they collaborated with each other to review manuscripts and write skill files to help our coding scaffold improve at doing research. The medical experts liked how quickly they could test different hypotheses and choose the best ones to continue working on Health research is pretty unique in that it is super important to be able to interpret noisy data and have taste to know which variables are important, which was still a significant value add from human researchers. Given that medicine is one of the oldest fields in human history and it is ultimately about humans, keeping a substantial human component in this type of research is probably important. In the near future I see AI primarily as a tool to facilitate human-led research. With the right scaffold I believe health researchers can be 10x or even 100x more productive using AI tools that have appropriate access to the wealth of data that the world has collected about human health One way to see this work is through the broader lens of AI for science. You can think of science as two parts: (1) collecting raw data about the world, and (2) synthesizing that raw data into understanding. In this project we already had existing data about the world and we aimed to accelerate part (2) using AI. Eventually, I imagine a world where we have such great AI tools that the time to synthesize raw data into understanding is driven towards zero. In that world, science will be substantially faster and most of the work will be focused on collecting the right raw data via physical experiments We did this in a pretty short amount of time (which was the point) and I think the future of scientific research will be more faster and more exciting than ever

We’re exploring how AI can accelerate health science research. While AI has made significant breakthroughs in formal sciences like mathematics, health data science depends on imperfect real-world data and consequential choices about cohorts, variables, and causal assumptions. To explore how AI can assist health researchers, we built a human-led research scaffold powered by Muse Spark 1.3 and Muse Wonder with access to 5M retrospective de-identified electronic health records and publicly available data. A cohort of medical experts used the system to conduct original research, accelerate hypothesis generation, code execution, and manuscript drafting while maintaining full authority over scientific decisions. Today, we’re sharing 10 research manuscripts produced through this exploration, spanning cardiology, oncology, infectious disease, emergency medicine, and more. Read more in our research blog: go.meta.me/3U06wvD

muse wonder full is SOTA on VCB 1-100, new coding eval out today

Muse Wonder Full tops VCB 1–100 which measures sustained, multi-turn software development in the real-world setting and mogs on cost!

that's just my face

hiiii 👋🏻

Excited to share Muse Wonder and Muse Wonder Full!

Brought to you by the world’s best agents research team. Our team puts an unbelievable amount of love, care and taste into developing Muse models. Benchmark scores alone won’t even tell how awesome our models are. If in doubt, keep calm and muse on yourself.

apparently visual coding is the social media meta 😅 3D demo PT sprint incoming

Some thoughts on the new Meta release today: - The models from MSL have improved quickly. They’re not just competing for Pareto optimal anymore, but driving for the frontier. - To me, the biggest unlock of these models is their token efficiency. They’re able to see significant performance increases while reasoning less and spending fewer tokens / turns. In Vibe Code Bench ,WF gets the same performance as MS 1.3 in half the number of steps. Smarter models think more efficiently. - Today, Muse Wonder Full is only available in in-app experiences (not via API). It may be that we see an even tighter coupling between models and first party applications. Or that this gating prevents nefarious actors from misusing the most capable intelligence. Not yet clear what the plan is. - MSL is making some serious progress on alignment. We’ll have more to share publicly here soon. Meta hasn’t been as vocal about safety, but to service Muse so widely they have needed to solve some fundamental alignment challenges. - Relatedly, we saw 0 refusals and Wonder did better than Wonder Full on our cyber capabilities benchmarks. This is itself very good! People should not expect their hard economically valuable work to be inhibited by these models. - The models have room for improvement on scientific discovery. They are strong executors, especially in code, but don’t have the intuition for which experiments to try. We’re still running our RSI Index, sharing results soon. If they aren’t able to maintain the frontier, this would be why. - META went up 3% today. I’ve said it before and I’ll say it again. As AI underpins a larger part of our economy, third party benchmarks will be the best value predictor of AI labs. The market for intelligence is rationalizing.