> ## Content Index > Fetch the complete content index at: https://www.implicator.ai/llms.txt > Use this file to discover other available public pages before exploring further. # Meta Releases 30B Open-Weight Muse Glimmer and Promises Spark 1.2 Weights - URL: https://www.implicator.ai/meta-releases-30b-open-weight-muse-glimmer-and-promises-spark-1-2-weights/ - Published: 2026-08-11T17:00:00.000Z - Updated: 2026-08-11T23:57:30.000Z - Description: Meta released Muse Glimmer under Apache 2.0 and promised to open Muse Spark 1.2. In May its own AI chief said Spark was not suitable for open sourcing, and Meta has not explained what changed. - Author: Marcus Schuler - Tags: AI News Meta released [Muse Glimmer](https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model?ref=implicator.ai), a 30-billion-parameter open-weight model, Monday. Glimmer distills Muse Spark to run local agents on a single high-end Mac or PC. Meta promised Muse Spark 1.2 weights in coming weeks; in an [interview released May 13](https://www.corememory.com/p/metas-ai-chief-alex-wang-muse-spark-ai-wars?ref=implicator.ai), Meta chief AI officer Alexandr Wang said Muse Spark was “not suitable for open sourcing.” What Changed - Meta released Muse Glimmer, a 30-billion-parameter open-weight model, on Monday under an Apache 2.0 license, and promised to publish the weights for its more capable Muse Spark 1.2 in the coming weeks. - In a Core Memory interview released May 13, Meta chief AI officer Alexandr Wang said Muse Spark was "not suitable for open sourcing." Meta has not publicly explained what changed. - Roughly 4-bit compression brings the model below 20 GB, targeting a 24 GB or 32 GB machine, which leaves a typical 8 GB or 16 GB laptop out of reach. - Apart from one independent hands-on run that measured about 12.2 tokens per second on a DGX Spark, the throughput and benchmark figures are Meta's own and have not been independently verified. AI-generated summary, reviewed by an editor. [More on our AI guidelines](https://www.implicator.ai/about/). ## What Meta released The [model card](https://huggingface.co/meta-models/Muse-Glimmer-30B?ref=implicator.ai) lists about 29.6 billion parameters across 52 layers at its August 2026 release, including a roughly 1.8-billion-parameter image encoder. Full-precision weights, two quantized versions, a speculative-decoding drafter and the perception encoder came under Apache 2.0\. The training data and training code remain private. Unlike Meta’s earlier Llama community license, Apache 2.0 carries no monthly-user cutoff. ## The hardware limit Full precision requires more than 55 GB of memory under Meta’s specifications. Its roughly 4-bit compression brings the language model below 20 GB, leaving room for working memory and the other components within a 24 GB or 32 GB machine. That excludes a typical 8 GB or 16 GB laptop at launch. Meta’s launch materials measured the drafter raising RTX 5090 output from 74.9 to 233.4 tokens per second at batch size one with greedy decoding. An [article published August 10](https://www.theregister.com/ai-and-ml/2026/08/10/zuck-rekindles-open-weights-llama-drama-with-muse-glimmer/5285666?ref=implicator.ai) says The Register ran Glimmer on a DGX Spark in Unsloth Studio and measured about 12.2 tokens per second without drafter support. The outlet estimated that machines lacking a dedicated graphics card with enough memory would manage roughly 6 to 14 tokens per second. Outside The Register’s run, Meta’s Glimmer throughput and benchmark measurements have not been independently verified. ## A smaller competitive entry Meta’s launch table scored Glimmer at 75.5 on MCP Atlas against Alibaba’s Qwen3.6-27B at 62.5\. It showed Qwen3.6-27B leading in computer use, terminal work and some coding tests. The Register’s Tobias Mann wrote that Glimmer “doesn’t exactly move the needle much on reclaiming American open weights superiority.” At release, Glimmer was far smaller than Moonshot AI’s 2.8-trillion-parameter Kimi K3 and Alibaba’s 2.4-trillion-parameter Qwen3.8-Max. By May 2026, Chinese open-weight models supplied roughly 61 percent of tokens consumed through OpenRouter, and Meta’s Llama no longer appeared in its rankings. Get Implicator.ai in your inbox Strategic AI news from San Francisco. No hype, no "AI will change everything" throat clearing. Just what moved, who won, and why it matters. Daily at 6am PST. Email address Subscribe Check your inbox. Click the link to confirm. No spam. Unsubscribe anytime. Spyglass writer [M.G. Siegler](https://spyglass.org/meta-open-ai-muse-glimmer/?ref=implicator.ai) called Meta’s insistence that it never abandoned open models “some gaslighting.” He wrote that Meta probably would say less about openness if the company led the AI market or if OpenAI and Anthropic were still “running away with the AI game.” ## The essay and the fund Zuckerberg published a 6,500-word essay Monday, “The Future Is for Everyone,” carrying the open-weight commitment. “The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic,” he wrote. He named neither OpenAI nor Anthropic; both keep leading model weights private. Know someone who'd find this useful? [✉️ Email it to a friend in one click](mailto:?subject=A%20newsletter%20I%20think%20you%27d%20like&body=This%20is%20one%20of%20maybe%20three%20newsletters%20I%20actually%20read.%20The%20rest%20just%20pile%20up%2C%20unread%2C%20judging%20me.%0A%0AAnd%20yes%2C%20this%20email%20mostly%20wrote%20itself%2C%20which%20is%20a%20little%20on%20the%20nose%20for%20an%20AI%20newsletter.%20Doesn%27t%20make%20it%20wrong.%20implicator.ai%20is%20good.%0A%0ASubscribe%20free%3A%20https%3A%2F%2Fwww.implicator.ai%2Fsubscribe%2F%3Futm%5Fsource%3Dnewsletter%26utm%5Fmedium%3Dforward%26utm%5Fcampaign%3Demail%5Fforward), or they can [subscribe free here](https://www.implicator.ai/subscribe/?utm%5Fsource=newsletter&utm%5Fmedium=forward&utm%5Fcampaign=forward%5Fto%5Fcolleague). The essay announced a $1 billion fund for Meta’s data-center communities. Meta plans up to $145 billion in capital spending this year and $600 billion by 2028; the fund is about 0.7 percent of the 2026 figure. Facing construction opposition, New York recently banned new large data centers for up to a year. ## The unexplained reversal In a Core Memory interview released May 13, Wang said Muse Spark had triggered checks for biological, chemical, cyber and loss-of-control risks. Meta’s [158-page safety report](https://ai.meta.com/static-resource/muse-spark-safety-and-preparedness-report/?ref=implicator.ai), published April 28, put Spark’s chemical and biological capabilities at “high risk” before safeguards and “moderate or lower” after them. Glimmer’s August model card places its risks at moderate or lower and says it falls outside Meta’s definition of frontier AI because it is generally less capable than Spark. It inferred moderate-or-lower cyber and loss-of-control ratings because Glimmer is broadly weaker than Muse Spark 1.0, which held the same designations. Meta has not publicly explained why Wang deemed Muse Spark unsuitable for open release in an interview released May 13 before Zuckerberg committed on August 10 to release Muse Spark 1.2’s weights. For Watermelon, the codename of Meta’s unreleased frontier model that the company has only alluded to, Siegler predicted: “My guess would be that if ‘Watermelon’ really is as good as the current frontier, the weights won’t be released but it will instead be used to distill some other Muse variant that Meta will release as an ‘open’ model.” Frequently Asked Questions What is Muse Glimmer? A 30-billion-parameter open-weight model Meta released on Monday under an Apache 2.0 license. It is distilled from Meta's closed Muse Spark model and built to run local AI agents on a single high-end Mac or PC. What hardware does it actually need? At full precision the model requires more than 55 GB of memory. Meta's roughly 4-bit compression brings the language model below 20 GB, leaving room for working memory and the other components inside a 24 GB or 32 GB machine. A typical 8 GB or 16 GB laptop is out of reach. How does it compare with rival models its size? In Meta's own launch table, Glimmer scored 75.5 on MCP Atlas against Alibaba's Qwen3.6-27B at 62.5\. The same table showed Qwen3.6-27B leading in computer use, terminal work and some coding tests. Is Muse Glimmer open source? The weights are open under Apache 2.0, along with two quantized versions, the speculative-decoding drafter and the perception encoder. Meta has not released the training data or the training code. Unlike the earlier Llama community license, Apache 2.0 carries no monthly-user cutoff. What did Meta say about Muse Spark 1.2? Zuckerberg committed to releasing its weights in the coming weeks. In an interview released May 13, Alexandr Wang had said Muse Spark triggered safety checks that made it unsuitable for open sourcing. Meta has not publicly explained the change. AI-generated summary, reviewed by an editor. [More on our AI guidelines](https://www.implicator.ai/about/). [Meta Opens Muse Glimmer as Zuckerberg Presses U.S. AI Policy ShiftOn Aug. 10, 2026, Meta released Muse Glimmer, a 30-billion-parameter model whose weights developers can download. It was built for agents that run on a Mac or PC. The release carried a policy argumenThe Implicator![](https://www.implicator.ai/content/images/2026/08/2026-08-10-06.57.25-meta-muse-glimmer-open-weight-policy@2x.webp)](https://www.implicator.ai/meta-muse-glimmer-zuckerberg-ai-policy/) [Moonshot Releases Kimi K3 Weights as Amodei Rejects Open-Weight BanMoonshot AI released the weights for its Kimi K3 model on Monday, a system with 2.8 trillion parameters. Developers can now download, modify and self-host K3, which the report described as the world'sThe Implicator![](https://www.implicator.ai/content/images/2026/07/2026-07-27-16.40.26-kimi_k3_weights@2x.webp)](https://www.implicator.ai/moonshot-releases-kimi-k3-weights-as-amodei-rejects-open-weight-ban-2/) [Germany's Soofi S AI Model Tops All Open-Source Rivals on German BenchmarksA German research consortium coordinated by the KI Bundesverband released Soofi S, an open-source German-English foundation model, this week, according to its pretraining report. In the team's tests, The Implicator![](https://www.implicator.ai/content/images/2026/07/2026-07-18-23.33.33-soofi-s-german-sovereign-ai-model@2x.webp)](https://www.implicator.ai/germanys-soofi-s-ai-model-tops-open-source-rivals/)