“Meta’s Muse Code is out of beta, with an SDK launching in developer preview for building agents on top of it and monthly subscription plans rolling out.”

#1 𝕏 Meta’s Muse Code exits beta with agent SDK preview #1 𝕏 Meta’s Muse Code is out of beta, with an SDK launching in developer preview for building agents on top of it and monthly subscription plans rolling out. Also covered by: @Alexandr Wang #2 𝕏 Google Research announced TimesFM-3, a state-of-the-art time series foundation model that enables multivariate forecasting in a single forward pass and is claimed to significantly outperform other forecasting models across major benchmarks.

“AI at Meta shared new evaluations and demos of Muse Spark 1.2, highlighting multimodal capabilities spanning visual-to-code generation, perception-to-physical action, and audio-visual understanding for video-heavy enterprise workflows.”

#2 𝕏 AI at Meta shared new evaluations and demos of Muse Spark 1.2, highlighting multimodal capabilities spanning visual-to-code generation, perception-to-physical action, and audio-visual understanding for video-heavy enterprise workflows. One demo shows the model parsing multimodal observations and calling tools to guide a robot through an unstructured environment to find a rubber duck. Also covered by: @AI at Meta , @Alexandr Wang

“Meta released Muse Glimmer, a 30 billion-parameter dense agentic model distilled from Muse Spark and licensed under Apache 2.0, with 4-bit quantization and DFlash speculative decoding for consumer-GPU use.”

#5 ▶️ Meta's new model wants "deep access" to your personal life... Fireship Meta released Muse Glimmer, a 30 billion-parameter dense agentic model distilled from Muse Spark and licensed under Apache 2.0, with 4-bit quantization and DFlash speculative decoding for consumer-GPU use. Muse Glimmer uses logit distillation from Meta's closed Muse Spark model; the transcript states it outperforms Gemma 4, performs comparably with Qwen 3.6, and has a 28% prompt-injection attack success rate versus 40% for Qwen. At full precision, Muse Glimmer requires more than 55 GB of memory; compressing its weights to approximately 4 bits reduces the model to just under 20 GB. A small draft model named DFlash generates token blocks for speculative decoding, after which Muse Glimmer verifies them in one pass; Meta reports a 3x speed increase on an NVIDIA 5090.

“"#4 𝕏 Sebastian Raschka recaps Meta’s release the previous day of Meta Muse Glimmer, an open-weight, dense 30B multimodal reasoning model with a 131k context window."”

#4 𝕏 Sebastian Raschka recaps Meta’s release the previous day of Meta Muse Glimmer, an open-weight, dense 30B multimodal reasoning model with a 131k context window. Its 32 query heads and 2 KV heads deliver a 52 KiB BF16 KV cache per token, making it memory-efficient and well suited to agentic workflows, though benchmarks conflict on whether it outperforms Qwen3.6. Also covered by: @Rowan Cheung , @Rowan Cheung

“AI at Meta released Muse Glimmer, an open-weight, 30B-parameter model optimized for local, always-on agent workflows on consumer hardware, including Macs and PCs with performant GPUs.”

Meta releases 30B Muse Glimmer for always-on local agents #1 𝕏 AI at Meta released Muse Glimmer, an open-weight, 30B-parameter model optimized for local, always-on agent workflows on consumer hardware, including Macs and PCs with performant GPUs. Its weights are available under the Apache 2.0 license. Also covered by: @Alexandr Wang , @AI at Meta , @Alexandr Wang , @NVIDIA AI

“Rowan Cheung says Mark Zuckerberg runs Meta’s superintelligence lab like a 50–100 person startup, personally recruiting every top AI researcher to keep the whole project in everyone’s head and avoid the outsized drag of underperformers.”

#16 𝕏 Rowan Cheung says Mark Zuckerberg runs Meta’s superintelligence lab like a 50–100 person startup, personally recruiting every top AI researcher to keep the whole project in everyone’s head and avoid the outsized drag of underperformers.

“AI at Meta is using SAM 3 and DINOv3 to automate image segmentation for Berkeley Lab’s SYNAPS-I project, slashing 3D volume labeling from a month of manual work to about 15 minutes.”

AI at Meta is using SAM 3 and DINOv3 to automate image segmentation for Berkeley Lab’s SYNAPS-I project, slashing 3D volume labeling from a month of manual work to about 15 minutes.

“Meta dropped Muse Spark 1.1, an agentic AI that outperforms OpenAI and Anthropic while massively undercutting their prices—proof that Zuckerberg’s tens-of-billions-dollar compute bet is paying off.”

#14 𝕏 Rowan Cheung : Meta dropped Muse Spark 1.1, an agentic AI that outperforms OpenAI and Anthropic while massively undercutting their prices—proof that Zuckerberg’s tens-of-billions-dollar compute bet is paying off. Meta’s stock has jumped over 10% since the release. #15 𝕏 Clem 🤗 highlights HuggingNews, an AI-curated feed by @ivan_bezdomny that cuts through the noise to surface the top AI stories.

“Meta released Muse Spark 1.1, the first Spark model to offer an API, with claimed improvements in agentic tool calling and computer use.”

The newsletter credits Meta with the Muse Spark 1.1 release and mentions related commentary from developers and AI observers.

“AI at Meta introduced Muse Image, its most advanced image-generation model offering precise edits, multi-reference composition, Instagram-powered context and agentic tool use via Muse Spark in the Meta AI app, Instagram Stories and WhatsApp.”

#1 𝕏 AI at Meta introduced Muse Image, its most advanced image-generation model offering precise edits, multi-reference composition, Instagram-powered context and agentic tool use via Muse Spark in the Meta AI app, Instagram Stories and WhatsApp. Also covered by: @AI at Meta , @Alexandr Wang