Today we're bringing specialized models trained using Muse Wonder to @Muse and Meta AI, giving users access to some of Muse Wonder's frontier capabilities for complex tasks at work and in everyday life. These specialized models include Muse Wonder's Active Memory and Active Focus, helping them stay aligned with your goals as tasks unfold: → Active Memory helps recognize when earlier information matters and distinguish what's still relevant from what's become stale → Active Focus helps recover from setbacks or find alternatives when constraints change In Muse, the improved models show a 2.4x reduction in forgetting information critical to your goals and a 67% improvement in getting unstuck as constraints evolve. That means less repeating yourself and fewer dead ends when plans change. In Meta AI, these improvements help it break complex problems into steps, ask sharper follow-up questions and adapt when you change direction. Read more on our research blog:

New on Open Math OS: Muse Spark Open by @AIatMeta 816B total parameters (42B active). 1M context window. Available now to all approved teams. Apply for Open Math OS now:

🧵We’re sharing ODAC25, the largest open dataset to date for Direct Air Capture materials ☁️ Published in @NatChemEng with @AIatMeta @GeorgiaTech, and leading academic partners.

Meta's @Muse answers questions very differently from ChatGPT or Anthropic. It cites more sources but uses far fewer words and tables, and doesn't try to ask a follow-up. This short-form content is designed for dealing with tasks rather than continuing a conversation, and frankly, it's far less exhausting to read. This speaks to a really well designed AI experience and its so different from the base LLMs. Great study by the AirOps team 👇 gg @alexandr_wang

How can Muse models help you build and scale? Join us tomorrow, Thursday, Oct 29 at 9:30am PT to explore coding and agentic use cases with Muse Wonder and Muse Spark 1.3, learn how each model fits into your stack, and get direct feedback from Meta experts in live Q&A. RSVP:

Muse Wonder is now available on @Cline, @Cloudflare, @Cognition, @Cursor_AI, @OpenRouter, @Oracle and @Vercel - with more partners coming soon. We look forward to seeing what the community builds with Muse Wonder! Start building 👉

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:

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:

American open weights are back. 🇺🇸 Huge credit to @finkd, @alexandr_wang and the Meta team for putting Muse Spark Open in developers’ hands. Muse Spark Open beats Nemotron 3 Ultra by a wide margin on agentic tasks. We’ve already added it to I for B200, with more configs to come. See what it costs to run, output throughput, etc. (1/2)🧵

Guys this is like my superbowl I waited so long for US open weight models to be relevant again. Before today the US models were still behind Chinese models from as early as Feb. 2026. Still more work to do! This cut the gap by at least half, in time (say 3 months)

Meta has released Muse Spark Open, an 816B (42B active) open-weights model that scores 39 on the Artificial Analysis Intelligence Index, making it the #1# US open-weights model Muse Spark Open (xhigh) scores 22 points above Muse Glimmer (high, 17), Meta's 30B dense open-weights model from August. It is also ahead of NVIDIA's Nemotron 3.5 Super VL (26), the next highest-scoring US open-weights model, which launched today @AIatMeta shared access with us ahead of public release for benchmarking. Congratulations to @AIatMeta, @finkd, and @alexandr_wang on the release! Key Takeaways: ➤ Open weights at the level of Meta's August flagship. Muse Spark Open scores within a point of Muse Spark 1.2 (xhigh, 40). It is stronger on AutomationBench-AA and has a lower hallucination rate, but weaker on GDPval-AA v2.1, CritPt and AA-Omniscience accuracy. Within Meta's current lineup, it sits 13 points behind Muse Wonder (max, 52) and 17 points behind Muse Wonder Full (max, 56) ➤ Token-efficient, at a moderate cost per task. Muse Spark Open uses 40k output tokens per Intelligence Index task, the fewest among the nine highest-scoring open-weights models: e.g. 16% fewer than Muse Spark 1.2 (xhigh, 48k) and 43% fewer than GLM-5.3 (max, 71k), its closest Chinese peer by size. Together AI, Fireworks and Baseten offer it at launch for $1.25/$4.25 per 1M input/output tokens ($0.15 cached input). At that pricing it costs $0.84 per Intelligence Index task. That is less than Muse Spark 1.2 ($0.97) at the same per-token prices and less than half of GLM-5.3 (max, $2.01), but about three times DeepSeek V4.1 Flash (max, $0.27) ➤ A relative weakness in agentic work compared to Chinese open-weights leaders. Among the nine highest-scoring open-weights models, Muse Spark Open ranks last on the four agentic evaluations in the index but fourth on the other six combined, and it leads all open-weights models on AA-Omniscience (25). GLM-5.3 (max), the closest Chinese model by size, owes all of its 6-point lead to the agentic evaluations ➤ Self-hostable on a single node, or two Mac Studios. BF16 weights are ~1.6TB. Meta's NVFP4 checkpoint quantizes only the 256 experts, cutting the weights to ~530GB. Meta says this runs on four Blackwell Ultra GPUs or eight H200s, or on two 512GB M5 Ultra Mac Studios using MLX. The architecture uses 62 layers with grouped-query attention (16 KV heads), 8 of 256 experts active per token, and a 1.8B-parameter Perception Encoder for vision Other model details: ➤ Parameters: 816B total, 42B active (MoE) ➤ Context window: 1M tokens ➤ License: Muse Model License, a custom license that requires "Built with Muse Spark Open" attribution and a separate agreement with Meta for Model-as-a-Service businesses with over $20M in annual revenue ➤ Input modalities: text, image, video and audio. Output: text ➤ Availability: weights on Hugging Face (BF16 and NVFP4), with API access via Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI

meta muse spark open beats 3 open-weight models – best pixel survivors game muse spark open builds the most playable game of the four for $0.59. same brief for all: a vampire survivors-style pixel game, plain canvas 2d, no libraries, no images, built in 4 steps cost #1# hy4 preview – $0.52 #2# inkling – $0.52 #3# muse spark open – $0.59 #4# step 5 preview – $0.60 time #1# inkling – 10m 53s #2# muse spark open – 19m 58s #3# step 5 preview – 20m 31s #4# hy4 preview – 52m 27s observations: • muse – the most game-like: hordes, damage numbers, level-ups, zero crashes • hy4 – slowest by far, but clean code at every step • step – crashed mid-build and needed an extra fix turn • inkling – fastest, prettiest graveyard, but its weapons deal zero damage follow @thehypedotnews for 24/7 ai news, analysis and breakdowns

Muse Wonder from Meta Superintelligence Labs is now available in Cursor! It scores 46.1% on CursorBench at default effort, the highest of any Muse model. Priced at $4/M input and $20/M output tokens.

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:

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: Download the weights:

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.

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

1/ muse spark open is the most capable US open-weight model to date. 816B total params, 42B active. 1M-token context. natively multimodal. built for coding and agentic work like long-horizon tasks, hard software engineering, multi-step reasoning. america! 🇺🇲