AI at Meta(@AIatMeta)

Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic t...

AI at Meta (@AIatMeta) — Mon, 19 Oct 2026 19:33:10 GMT

Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic tasks on consumer hardware.

It runs at practical speeds on a Mac or PC with a single consumer GPU at 4-bit precision, and leads comparably-sized open models on tool use and multi-step planning.

Read about the model architecture, distillation recipe, quantization and more: go.meta.me/4Bcvxol

💬77🔄97❤️755👀72716📊216

⚡ Powered by xgo.ing

We trained our Muse foundational models, including Muse Wonder, at Prometheus, our gigawatt-scale su...

AI at Meta (@AIatMeta) — Mon, 19 Oct 2026 18:14:17 GMT

We trained our Muse foundational models, including Muse Wonder, at Prometheus, our gigawatt-scale supercluster in New Albany, Ohio.

Across a supercluster with hundreds of thousands of GPUs, even one underperforming host can slow an entire training run. We engineered Prometheus for resilience at scale, achieving 85%+ goodput on pretraining, 90%+ on reinforcement learning and ~47K GPU hours mean time between interruptions during Muse Wonder training.

Take a look at the tech powering Prometheus: go.meta.me/4yXXApZ

Your browser does not support the video tag.

🔗 View on Twitter

💬110🔄189❤️2426👀155653📊414

⚡ Powered by xgo.ing

We’re proud to share that Meta Omnilingual Automatic Speech Recognition (ASR) has been named one of ...

AI at Meta (@AIatMeta) — Thu, 15 Oct 2026 22:32:35 GMT

We’re proud to share that Meta Omnilingual Automatic Speech Recognition (ASR) has been named one of @TIME’s Best Inventions of 2026!

Meta Omnilingual ASR is a suite of models providing ASR capabilities for over 1,600 languages, including 500 low-coverage languages never before served by any ASR system.

While most ASR systems focus on a limited set of languages that are well-represented on the internet, this release marked a major step toward building a truly universal transcription system.

Learn more about it here: time.com/collection/bes…

Your browser does not support the video tag.

🔗 View on Twitter

💬74🔄72❤️679👀46939📊167

⚡ Powered by xgo.ing

To help make advanced models more accessible, Muse Spark Open is optimized to run on two Apple M5 Ul...

AI at Meta (@AIatMeta) — Thu, 15 Oct 2026 17:08:05 GMT

To help make advanced models more accessible, Muse Spark Open is optimized to run on two Apple M5 Ultra Mac Studios with the latest MLX framework.

It also runs on leading inference engines like SGLang, vLLM, TensorRT-LLM, Hugging Face Transformers, llama.cpp, and Unsloth with optimizations for AMD, Apple, Intel, and NVIDIA silicon.

Developers can further customize the model with PyTorch, including TorchTitan for fine-tuning.

💬26🔄38❤️780👀109292📊120

⚡ Powered by xgo.ing

Muse Spark Open is available today through API platforms at Baseten, Fireworks AI, Groq, Modal, Nebi...

AI at Meta (@AIatMeta) — Thu, 15 Oct 2026 17:08:05 GMT

Muse Spark Open is available today through API platforms at Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI.

Download the weights on Amazon SageMaker AI, Google Cloud Model Garden, Hugging Face, and Microsoft Foundry Models.

Weights are released under the Muse Model License, with limited commercial terms that let many developers use the model for free.

Learn more on our research blog: go.meta.me/musesparkopen

Download the weights: huggingface.co/collections/me…

💬6🔄25❤️434👀59594📊63

⚡ Powered by xgo.ing

Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for cod...

AI at Meta (@AIatMeta) — Thu, 15 Oct 2026 17:08:04 GMT

Today we’re releasing Muse Spark Open, the most capable US open-weight model to date.

Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.

Key Model Specs:
→ 816B total / 42B active parameters
→ 1M token context window
→ Natively multimodal across text, image, video, audio

💬473🔄1222❤️11516👀1218276📊2212

⚡ Powered by xgo.ing

Muse Wonder and Muse Wonder Full deliver SOTA performance on personal and professional agentic tasks...

AI at Meta (@AIatMeta) — Wed, 14 Oct 2026 23:46:15 GMT

Muse Wonder and Muse Wonder Full deliver SOTA performance on personal and professional agentic tasks. 

Learn more about the models on our researcher blog: go.meta.me/musewonder

Your browser does not support the video tag.

🔗 View on Twitter

💬60🔄54❤️546👀53547📊123

⚡ Powered by xgo.ing

Watch the full convo with @finkd, @alexandr_wang and @collision: https://t.co/AJ9GHXjB2d

AI at Meta (@AIatMeta) — Wed, 14 Oct 2026 23:46:15 GMT

Watch the full convo with @finkd, @alexandr_wang and @collision:

x.com/collision/stat…

John Collison@collision

Cheeky Pint is back! Mark Zuckerberg @finkd and Alexandr Wang @alexandr_wang came to the pub to discuss their new frontier model, Muse Wonder.

00:24 Muse’s breakout use cases
07:28 The model is the product
14:35 Muse Wonder
24:56 How frontier models are built
48:26 Betting $100B
07:21 Agentic commerce
19:34 OpenRouter 
20:06 Beyond the smartphone
27:00 Funding
34:06 Why AI won’t be winner-take-all
40:46 Meta Superintelligence Labs (MSL)
08:11 Working for Mark Zuckerberg

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬1🔄0❤️19👀11464📊4

⚡ Powered by xgo.ing

We’re exploring how AI can accelerate health science research. While AI has made significant breakt...

AI at Meta (@AIatMeta) — Wed, 14 Oct 2026 18:44:49 GMT

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

Your browser does not support the video tag.

🔗 View on Twitter

💬73🔄99❤️717👀213325📊201

⚡ Powered by xgo.ing

@Kinzf2121 Absolutely, Muse Spark open weights are coming soon!

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 18:41:59 GMT

@Kinzf2121 Absolutely, Muse Spark open weights are coming soon!

💬1🔄6❤️42👀43700📊9

⚡ Powered by xgo.ing

@TheINAOG The open weights for Muse Spark are coming soon, Muse Wonder models are large models that ...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:43:53 GMT

@TheINAOG The open weights for Muse Spark are coming soon, Muse Wonder models are large models that are too big for many people to use, in addition to potential safety concerns.

💬45🔄47❤️684👀68650📊128

⚡ Powered by xgo.ing

From rebuilding our foundation to pushing the frontier in 14 months 🚀

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:32:22 GMT

From rebuilding our foundation to pushing the frontier in 14 months 🚀

💬92🔄105❤️2139👀129032📊309

⚡ Powered by xgo.ing

To quantify how the models handle unfamiliar tasks, we compared test-time compute scaling on tasks t...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:29:29 GMT

To quantify how the models handle unfamiliar tasks, we compared test-time compute scaling on tasks trained on during post-training against other tasks.

Both models improve similarly out-of-distribution, but given the same in-distribution score, Muse Wonder Full performs better on unfamiliar tasks than the Muse Spark 1.3 trend predicts. We attribute this upward shift to a stronger base model, which carries post-training gains over more effectively to tasks we didn't anticipate.

💬3🔄0❤️11👀3960📊4

⚡ Powered by xgo.ing

Delivering personal superintelligence requires a significant investment in alignment and safety, and...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:02:23 GMT

Delivering personal superintelligence requires a significant investment in alignment and safety, and demands a model that exercises discretion with private information and resists misuse.

We trained the Muse Wonder family to lead on both dimensions, with robustness to adversarial instructions and careful handling of private information.

💬6🔄5❤️81👀6436📊14

⚡ Powered by xgo.ing

We hope that Muse Wonder and Muse Wonder Full give more people the power to pursue their dreams and ...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:02:23 GMT

We hope that Muse Wonder and Muse Wonder Full give more people the power to pursue their dreams and build the world they want.

Learn more in our research blog: go.meta.me/musewonder

💬5🔄1❤️60👀5316📊10

⚡ Powered by xgo.ing

Two things make Muse Wonder suited for long-running, multi-step work an agent faces every day: -> A...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:02:22 GMT

Two things make Muse Wonder suited for long-running, multi-step work an agent faces every day:

-> Active Memory carries your context across conversations and surfaces it when it's relevant.

-> Active Focus keeps the models working toward your goal when a task gets challenging, trying alternative approaches, working through blockers, and following through to completion.

💬16🔄7❤️88👀17568📊24

⚡ Powered by xgo.ing

The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and prof...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:02:21 GMT

The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and professional agents, memory, and coding.

Muse Wonder and Muse Wonder Full apply specialized knowledge to problems people encounter at work and in daily life, from understanding a lease to making sense of financial or health information. Our evaluations include tasks in law, finance, and health that test this ability.

💬9🔄8❤️117👀8205📊23

⚡ Powered by xgo.ing

We’re excited to introduce the Muse Wonder series with Muse Wonder and Muse Wonder Full, our most ca...

AI at Meta (@AIatMeta) — Mon, 12 Oct 2026 17:02:20 GMT

We’re excited to introduce the Muse Wonder series with Muse Wonder and Muse Wonder Full, our most capable models yet. Both are frontier models, built to deliver state-of-the-art performance for personal and professional agentic tasks.

Muse Wonder is available today via the Meta Model API and Muse Code. Muse Wonder Full is available today for Muse Code subscribers. Muse Wonder will also soon power experiences in our personal agent product, Muse, rolling out first to subscribers.

Your browser does not support the video tag.

🔗 View on Twitter

💬208🔄235❤️2190👀307174📊534

⚡ Powered by xgo.ing

6️⃣ Non-Associative Algebra: The team worked with Muse Spark to find an exception to a proposed rule...

AI at Meta (@AIatMeta) — Fri, 02 Oct 2026 19:11:35 GMT

6️⃣ Non-Associative Algebra: The team worked with Muse Spark to find an exception to a proposed rule about mathematical structures inspired by biology. They went further by developing an alternative characterization, which the researchers checked and refined.

Read the paper: ai.meta.com/research/publi…

💬5🔄1❤️63👀15263📊11

⚡ Powered by xgo.ing

1️⃣ Probability: Mathematicians worked with Muse Spark to answer a question about fitting random po...

AI at Meta (@AIatMeta) — Fri, 02 Oct 2026 19:11:31 GMT

1️⃣ Probability:  Mathematicians worked with Muse Spark to answer a question about fitting random points onto the surface of a stretched sphere. For the setting studied, they proved a sharp cutoff between when an exact fit is likely and when it is unlikely.

Read the paper: ai.meta.com/research/publi…

💬11🔄8❤️157👀96390📊33

⚡ Powered by xgo.ing

Following gold-medal-level performance from our AI models across five competitions in mathematics, p...

AI at Meta (@AIatMeta) — Fri, 02 Oct 2026 19:11:30 GMT

Following gold-medal-level performance from our AI models across five competitions in mathematics, physics, and chemistry, we asked a harder question: can AI contribute when a problem is genuinely open and without an existing solution path?

Over the past several months, mathematicians worked with Muse Spark 1.1 and Muse Spark 1.2 in Thinking Mode through the regular meta.ai chat interface, with no custom research scaffold, to find solutions to such problems.

Our goal wasn't to mass-produce papers, but empower researchers. Every collaboration followed the same principles: mathematicians guided the research, a second group of mathematicians then reviewed the work, each paper marks which passages were drafted primarily by humans or AI, and each credits the prior research it builds on. Where other teams independently announced solutions to the same problems, we acknowledge their work as well.

Today, we're sharing six papers from that collaboration.

🧵👇

💬112🔄139❤️1204👀390262📊296

⚡ Powered by xgo.ing

In the past 6 months... - Muse Spark - Muse Image - Muse Video - Muse Spark 1.1 - Meta Model API - ...

AI at Meta (@AIatMeta) — Mon, 28 Sep 2026 14:30:01 GMT

In the past 6 months...

- Muse Spark
- Muse Image
- Muse Video
- Muse Spark 1.1
- Meta Model API
- Muse Glimmer
- Muse Spark 1.2
- Muse Code
- Muse Spark 1.3
- Muse

...and we’re just getting started. 🚀

Your browser does not support the video tag.

🔗 View on Twitter

💬208🔄147❤️2823👀190440📊488

⚡ Powered by xgo.ing

We put Muse Realtime Avatar head-to-head with two leading commercial avatar systems in their own liv...

AI at Meta (@AIatMeta) — Thu, 24 Sep 2026 05:43:22 GMT

We put Muse Realtime Avatar head-to-head with two leading commercial avatar systems in their own live-call products. Raters held 2–3 minute conversations with matched avatar identities, then compared visual quality, sync, character consistency, and mannerisms.

Muse Realtime Avatar came out ahead on overall preference.

💬7🔄3❤️24👀10165📊9

⚡ Powered by xgo.ing

Read our research blog to learn more about Muse Realtime Avatar: https://t.co/pVq9K0fxCG

AI at Meta (@AIatMeta) — Thu, 24 Sep 2026 05:43:22 GMT

Read our research blog to learn more about Muse Realtime Avatar: research.meta.ai/blog/bringing-…

💬4🔄2❤️22👀9553📊7

⚡ Powered by xgo.ing

@finkd just unveiled Muse Realtime Avatar, our real-time embodiment technology that turns Muse Realt...

AI at Meta (@AIatMeta) — Thu, 24 Sep 2026 05:43:20 GMT

@finkd just unveiled Muse Realtime Avatar, our real-time embodiment technology that turns Muse Realtime Voice into expressive, interactive avatars.

Muse Realtime Avatar enables an entirely new range of interactions in @Muse, starting with real-time conversations.

🧵👇#MetaConnectct

Your browser does not support the video tag.

🔗 View on Twitter

💬74🔄56❤️493👀114911📊140

⚡ Powered by xgo.ing

@Muse for Mac, rolling out today. 👇

AI at Meta (@AIatMeta) — Thu, 17 Sep 2026 22:33:21 GMT

@Muse for Mac, rolling out today. 👇

💬91🔄28❤️495👀68060📊131

⚡ Powered by xgo.ing

Today we launched https://t.co/KLJSHpx0pZ, a personal AI agent, and published a deep dive on how we ...

AI at Meta (@AIatMeta) — Tue, 08 Sep 2026 19:27:22 GMT

Today we launched muse.ai, a personal AI agent, and published a deep dive on how we built safety into its system.

An agent that gets to know you over time necessarily holds a lot of context about you. That's the source of its usefulness and the reason we built Muse to be secure, safe, and private.

Read the full deep dive here: security.muse.ai

Your browser does not support the video tag.

🔗 View on Twitter

💬142🔄131❤️1235👀130593📊295

⚡ Powered by xgo.ing

Today we launched https://t.co/wErICEYbFU, a personal AI agent, and published a deep dive on how we ...

AI at Meta (@AIatMeta) — Tue, 08 Sep 2026 19:18:32 GMT

Today we launched Muse.ai, a personal AI agent, and published a deep dive on how we built safety into its system.

An agent that gets to know you over time necessarily holds a lot of context about you. That's the source of its usefulness and the reason we built Muse to be secure, safe, and private.

Read the full deep dive here: security.muse.ai

Your browser does not support the video tag.

🔗 View on Twitter

💬6🔄5❤️51👀3068📊13

⚡ Powered by xgo.ing

Introducing Muse, a personal agent that gets things done for you, powered by Muse Spark 1.3. Get an...

AI at Meta (@AIatMeta) — Tue, 08 Sep 2026 19:07:38 GMT

Introducing Muse, a personal agent that gets things done for you, powered by Muse Spark 1.3.

Get an inside look at how we built Muse and what it can do for you: introducing.muse.ai

Your browser does not support the video tag.

🔗 View on Twitter

💬227🔄328❤️3129👀555502📊714

⚡ Powered by xgo.ing

While the Kaggle competition demonstrated AIRA₃’s capabilities in a specific domain, the system itse...

AI at Meta (@AIatMeta) — Sat, 05 Sep 2026 16:17:40 GMT

While the Kaggle competition demonstrated AIRA₃’s capabilities in a specific domain, the system itself can  generalize across distinct domains: changing only the task specification. In an internal benchmark, AIRA₃ achieved a 27% latency reduction on production GPU kernels, and gold-level performance in another Kaggle competition translating 4,000-year-old Akkadian clay tablets into English.

We're early, and hard problems are still ahead of us. But we believe a system that compounds its own knowledge is the right bet. As we continue to develop and scale, we’re excited about its potential to accelerate AI research and unlock recursive self-improvement.

💬6🔄4❤️83👀15329📊15

⚡ Powered by xgo.ing

Rather than relying on a central controller, AIRA₃ runs many long-running agents (pairs of models + ...

AI at Meta (@AIatMeta) — Sat, 05 Sep 2026 16:17:39 GMT

Rather than relying on a central controller, AIRA₃ runs many long-running agents (pairs of models + coding harnesses) in their own isolated environments and coordinates asynchronously through two shared substrates:

1️⃣ a forum for sharing hypotheses and findings
2️⃣ a shared filesystem for solution artifacts

The individual agents collaborate, share discoveries, and build on each other’s work. As the graph below demonstrates, the system uses compute to compound knowledge and drive performance gains over time. Search strategies emerge dynamically as each agent decides which discoveries to build upon.

💬9🔄5❤️119👀22247📊25

⚡ Powered by xgo.ing

We entered AIRA₃ with an ensemble of models in the live competition, and also assessed it with sever...

AI at Meta (@AIatMeta) — Sat, 05 Sep 2026 16:17:38 GMT

We entered AIRA₃ with an ensemble of models in the live competition, and also assessed it with several others post-hoc. The 8th ranked gold medal entry ensemble was a combination of GPT 5.5 (w/ OpenCode) + Claude 4.8 (w/ ClaudeCode).

Post-hoc we assessed with Muse Spark 1.2 (w/ MuseCode), which also performed at a Gold Medal level, as well as Muse Spark 1.1 (w/ OpenCode) and GLM 5.2 (w/ OpenCode), both of which achieved Silver Medal level performance. The post-hoc submissions were also graded externally on the same private test set as those made during the live competition.

💬22🔄9❤️132👀24078📊35

⚡ Powered by xgo.ing

As a test of our progress to advance the frontier of AI research, in June we entered the next genera...

AI at Meta (@AIatMeta) — Sat, 05 Sep 2026 16:17:37 GMT

As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set.

AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools.

We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.

Your browser does not support the video tag.

🔗 View on Twitter

💬110🔄214❤️1982👀319502📊453

⚡ Powered by xgo.ing

Muse Spark 1.3 with max reasoning is now available on Muse Code and Meta Model API 👇

AI at Meta (@AIatMeta) — Fri, 04 Sep 2026 18:20:21 GMT

Muse Spark 1.3 with max reasoning is now available on Muse Code and Meta Model API 👇

Alexandr Wang@alexandr_wang

1/ we just publicly released Muse Spark 1.3 max!

we see significantly stronger coding and agentic performance on muse spark 1.3 max, so would strongly recommend trying it out even if you've already tried muse spark 1.3 high or muse spark 1.3 xhigh.

Your browser does not support the video tag.

🔗 View on Twitter

Your browser does not support the video tag.

🔗 View on Twitter

Your browser does not support the video tag.

🔗 View on Twitter

Your browser does not support the video tag.

🔗 View on Twitter

🔗 View Quoted Tweet

💬60🔄44❤️788👀59056📊108

⚡ Powered by xgo.ing

Muse Spark 1.3 is rolling out today in Muse Code and Meta Model API at https://t.co/EMi0nXEvtz, with...

AI at Meta (@AIatMeta) — Wed, 02 Sep 2026 19:36:20 GMT

Muse Spark 1.3 is rolling out today in Muse Code and Meta Model API at dev.meta.ai, with max reasoning coming soon after we finish safety testing.

Learn more: go.meta.me/46IsHsV

💬6🔄4❤️135👀17277📊18

⚡ Powered by xgo.ing

Stay tuned for more updates soon, including bigger models, Muse Spark open weights, and more. 🚀

AI at Meta (@AIatMeta) — Wed, 02 Sep 2026 19:36:20 GMT

Stay tuned for more updates soon, including bigger models, Muse Spark open weights, and more. 🚀

💬3🔄6❤️159👀19833📊19

⚡ Powered by xgo.ing

We’re excited to release Muse Spark 1.3 with improved performance on agentic and coding tasks, and a...

AI at Meta (@AIatMeta) — Wed, 02 Sep 2026 19:36:19 GMT

We’re excited to release Muse Spark 1.3 with improved performance on agentic and coding tasks, and a focus on real-world usability.

Key capabilities:
→ Sustains longer-horizon work across multiple workflows in a single thread
→ More actively collaborates with users: it asks clarifying questions, flags when it's stuck, confirms before consequential actions
→ Better calibrated on its own limits instead of hallucinating outcomes
→ ~20% fewer tool calls and ~25% fewer tokens vs. Muse Spark 1.2 in internal comparisons

💬148🔄293❤️3120👀434663📊528

⚡ Powered by xgo.ing

Muse Voice Transcribe is available today via Meta Model API, Meta AI for Mac, and Muse Code. Learn ...

AI at Meta (@AIatMeta) — Tue, 01 Sep 2026 17:26:10 GMT

Muse Voice Transcribe is available today via Meta Model API, Meta AI for Mac, and Muse Code.

Learn more: go.meta.me/4xISnl2

💬3🔄4❤️41👀8530📊9

⚡ Powered by xgo.ing

With adaptive delay, Muse Voice Transcribe achieves the pareto frontier on speed-accuracy trade-off ...

AI at Meta (@AIatMeta) — Tue, 01 Sep 2026 17:26:09 GMT

With adaptive delay, Muse Voice Transcribe achieves the pareto frontier on speed-accuracy trade-off measured by time to final transcription.

💬3🔄5❤️52👀9906📊9

⚡ Powered by xgo.ing

Introducing Muse Voice Transcribe, the first real-time audio perception model from Meta Superintelli...

AI at Meta (@AIatMeta) — Tue, 01 Sep 2026 17:26:08 GMT

Introducing Muse Voice Transcribe, the first real-time audio perception model from Meta Superintelligence Labs.

Muse Voice Transcribe delivers real-time streaming ASR, diarization with 20+ speakers, and endpointing. It’s multilingual with seamless code-switching and improves accuracy with language, keyword, and context biasing.

The model ranks first on @ArtificialAnlys streaming speech-to-text and on public diarization benchmarks.

💬77🔄184❤️1503👀142322📊320

⚡ Powered by xgo.ing

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Plywood Whale: https://t.co/q5dATH...

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 18:25:07 GMT

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Plywood Whale: research.meta.ai/wild-artifact-…

💬1🔄2❤️23👀8186📊5

⚡ Powered by xgo.ing

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Cat Mesh: https://t.co/loY5g46x1V

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 18:25:06 GMT

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Cat Mesh: research.meta.ai/wild-artifact-…

💬4🔄6❤️65👀11385📊12

⚡ Powered by xgo.ing

Today we’re also previewing WildArtifactBench, an internal evaluation framework designed to assess a...

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 18:25:05 GMT

Today we’re also previewing WildArtifactBench, an internal evaluation framework designed to assess agents on complex, real-world tasks across diverse deliverable formats.

By using win rates and Elo scores from human and agentic preference judges rather than strict ground-truth rubrics, it expands task coverage across practical multimodal workflows.

We’re releasing 10 tasks from WildArtifactBench as a step forward in our ability to measure the real practical utility delivered by multimodal agents: research.meta.ai/wild-artifact-…

💬92🔄59❤️633👀151508📊170

⚡ Powered by xgo.ing

Muse Spark brings spatial intelligence to robotics. A specialized variant of Muse Spark acts as the ...

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 17:06:57 GMT

Muse Spark brings spatial intelligence to robotics. A specialized variant of Muse Spark acts as the robot brain and orchestrator: it takes a user instruction, decodes tool calls, observes the results, and loops until the task is complete.

As shown in this demo, Muse Spark 1.2 can plan sub-tasks for a bimanual robot tidying a desk, distinguishing a hair brush from a makeup brush and placing the lipstick in a drawer.

Your browser does not support the video tag.

🔗 View on Twitter

💬5🔄5❤️32👀6546📊9

⚡ Powered by xgo.ing

Internally, Muse Spark is deployed in several areas including media generation. For example, the mod...

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 17:06:57 GMT

Internally, Muse Spark is deployed in several areas including media generation. For example, the model works with Muse Image for agentic media generation and produces detailed captions as training data for Muse Image and Muse Video. Muse Spark is also able to translate raw text, image, and video content into signals and insights that can be used in downstream applications.

See more Muse Spark 1.2 evals and demos here: go.meta.me/multimodal

Your browser does not support the video tag.

🔗 View on Twitter

💬3🔄3❤️23👀6140📊6

⚡ Powered by xgo.ing

Muse Spark 1.2 supports a broad range of multimodal tasks, from turning visuals into working code to...

AI at Meta (@AIatMeta) — Thu, 20 Aug 2026 17:06:55 GMT

Muse Spark 1.2 supports a broad range of multimodal tasks, from turning visuals into working code to translating perception into physical action. It also brings robust audio-visual understanding to enable video-heavy workflows common in real-world enterprise use.

Today, we’re sharing new evals and demos that illustrate the breadth of the model’s visual understanding and reasoning capabilities.

Let’s start with a demo that shows how Muse Spark parses multimodal observations and calls tools to guide a robot to navigate in an unstructured environment to find a rubber duck.

🧵👇

Your browser does not support the video tag.

🔗 View on Twitter

💬48🔄75❤️531👀58244📊121

⚡ Powered by xgo.ing

🔗 Download Muse Glimmer on @huggingface: https://t.co/s7Lzb8MqCG 🔗 Read the technical blog: https...

AI at Meta (@AIatMeta) — Mon, 10 Aug 2026 10:13:36 GMT

🔗 Download Muse Glimmer on @huggingface:huggingface.co/meta-modelsG

🔗 Read the technical bloggo.meta.me/museglimmerbc

🔗 Find resourcedeveloper.meta.com/ai/models/muse…FwU

💬13🔄74❤️572👀83356📊121

⚡ Powered by xgo.ing

Muse Glimmer can complete multi-step agentic tasks end-to-end from a single natural language prompt....

AI at Meta (@AIatMeta) — Mon, 10 Aug 2026 10:13:35 GMT

Muse Glimmer can complete multi-step agentic tasks end-to-end from a single natural language prompt.

In this demo, it autonomously discovers a local Home Assistant instance via network tool calls, queries device APIs, writes a responsive HTML/CSS/JS dashboard from scratch, and deploys a local server for verification.

Your browser does not support the video tag.

🔗 View on Twitter

💬9🔄34❤️427👀52604📊66

⚡ Powered by xgo.ing

For a local agent to be practical, generation latency must be low enough to maintain workflow contin...

AI at Meta (@AIatMeta) — Mon, 10 Aug 2026 10:13:35 GMT

For a local agent to be practical, generation latency must be low enough to maintain workflow continuity.

To run Muse Glimmer on consumer hardware without degrading quality, we used quantization to shrink the language model to under 20GB and a lightweight DFlash drafter model to accelerate token generation. As a result, Muse Glimmer is fast enough for fluid conversation and real-time agent interaction, all running entirely on your device.

💬7🔄25❤️458👀49798📊64

AI at Meta(@AIatMeta)

Today we published a technical report on Muse Glimmer, our 30B open-weight model built for agentic t...

We trained our Muse foundational models, including Muse Wonder, at Prometheus, our gigawatt-scale su...

We’re proud to share that Meta Omnilingual Automatic Speech Recognition (ASR) has been named one of ...

To help make advanced models more accessible, Muse Spark Open is optimized to run on two Apple M5 Ul...

Muse Spark Open is available today through API platforms at Baseten, Fireworks AI, Groq, Modal, Nebi...

Today we’re releasing Muse Spark Open, the most capable US open-weight model to date. Built for cod...

Muse Wonder and Muse Wonder Full deliver SOTA performance on personal and professional agentic tasks...

Watch the full convo with @finkd, @alexandr_wang and @collision: https://t.co/AJ9GHXjB2d

We’re exploring how AI can accelerate health science research. While AI has made significant breakt...

@Kinzf2121 Absolutely, Muse Spark open weights are coming soon!

@TheINAOG The open weights for Muse Spark are coming soon, Muse Wonder models are large models that ...

From rebuilding our foundation to pushing the frontier in 14 months 🚀

To quantify how the models handle unfamiliar tasks, we compared test-time compute scaling on tasks t...

Delivering personal superintelligence requires a significant investment in alignment and safety, and...

We hope that Muse Wonder and Muse Wonder Full give more people the power to pursue their dreams and ...

Two things make Muse Wonder suited for long-running, multi-step work an agent faces every day: -> A...

The Muse Wonder series achieves state-of-the-art results across key benchmarks for personal and prof...

We’re excited to introduce the Muse Wonder series with Muse Wonder and Muse Wonder Full, our most ca...

6️⃣ Non-Associative Algebra: The team worked with Muse Spark to find an exception to a proposed rule...

Read more on our research blog: https://t.co/J0e9snbE8Q

1️⃣ Probability: Mathematicians worked with Muse Spark to answer a question about fitting random po...

Following gold-medal-level performance from our AI models across five competitions in mathematics, p...

In the past 6 months... - Muse Spark - Muse Image - Muse Video - Muse Spark 1.1 - Meta Model API - ...

We put Muse Realtime Avatar head-to-head with two leading commercial avatar systems in their own liv...

Read our research blog to learn more about Muse Realtime Avatar: https://t.co/pVq9K0fxCG

@finkd just unveiled Muse Realtime Avatar, our real-time embodiment technology that turns Muse Realt...

@Muse for Mac, rolling out today. 👇

Today we launched https://t.co/KLJSHpx0pZ, a personal AI agent, and published a deep dive on how we ...

Today we launched https://t.co/wErICEYbFU, a personal AI agent, and published a deep dive on how we ...

Introducing Muse, a personal agent that gets things done for you, powered by Muse Spark 1.3. Get an...

While the Kaggle competition demonstrated AIRA₃’s capabilities in a specific domain, the system itse...

Rather than relying on a central controller, AIRA₃ runs many long-running agents (pairs of models + ...

We entered AIRA₃ with an ensemble of models in the live competition, and also assessed it with sever...

As a test of our progress to advance the frontier of AI research, in June we entered the next genera...

Muse Spark 1.3 with max reasoning is now available on Muse Code and Meta Model API 👇

Muse Spark 1.3 is rolling out today in Muse Code and Meta Model API at https://t.co/EMi0nXEvtz, with...

Stay tuned for more updates soon, including bigger models, Muse Spark open weights, and more. 🚀

We’re excited to release Muse Spark 1.3 with improved performance on agentic and coding tasks, and a...

Muse Voice Transcribe is available today via Meta Model API, Meta AI for Mac, and Muse Code. Learn ...

With adaptive delay, Muse Voice Transcribe achieves the pareto frontier on speed-accuracy trade-off ...

Introducing Muse Voice Transcribe, the first real-time audio perception model from Meta Superintelli...

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Plywood Whale: https://t.co/q5dATH...

Here are the generated artifacts of Muse Spark 1.1 and 1.2 on the Cat Mesh: https://t.co/loY5g46x1V

Today we’re also previewing WildArtifactBench, an internal evaluation framework designed to assess a...

Muse Spark brings spatial intelligence to robotics. A specialized variant of Muse Spark acts as the ...

Internally, Muse Spark is deployed in several areas including media generation. For example, the mod...

Muse Spark 1.2 supports a broad range of multimodal tasks, from turning visuals into working code to...

🔗 Download Muse Glimmer on @huggingface: https://t.co/s7Lzb8MqCG 🔗 Read the technical blog: https...

Muse Glimmer can complete multi-step agentic tasks end-to-end from a single natural language prompt....

For a local agent to be practical, generation latency must be low enough to maintain workflow contin...