Meta Superintelligence Labs has released Muse Ember, its smallest and most cost-efficient model so far, aimed at repetitive work inside AI agents rather than serving as a flagship general-purpose system. The company says it is built for tasks such as calling tools, completing coding subtasks, routing requests, extracting information, and parsing long documents. Meta highlights a price of 10 cents per million input tokens and uses Artificial Analysis’s composite benchmark to argue that Ember offers a stronger cost-and-performance balance than models such as Anthropic’s Haiku and OpenAI’s Luna, while also competing on price with open models including DeepSeek V4.1 Flash and GLM 5.3 Flash.
Ember is intended to handle high volumes of routine tasks alongside a more capable model: for example, Muse Wonder could establish how a customer-support issue should be handled, while Ember applies those rules across many cases. Meta’s “planner and executor” approach is designed to reduce the expense of deploying agents at scale. Ember also accepts text, images, and audio, enabling agents to process screenshots, charts, scanned documents, voice memos, meeting recordings, and video without switching to separate models. It is the fourth Muse model, following Spark, the open-source Glimmer, and the frontier Muse Wonder; Meta’s Muse Agent consumer product launched in September.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Researchers tested 16 LLMs on all 55 TIS-100 puzzles, giving each one attempt and time to refine its code. The models found 133 new performance tradeoffs across 18 puzzles and set three outright cycle records, though human players improved on most solutions within days. (tis-100.wincon.dev)
Australia is considering a new criminal offence for AI companies that fail to take reasonable steps to prevent harm caused by their models. The government is also weighing an AI safety watchdog with power to impose significant fines, following a hack of a Medicare statistics portal involving OpenAI. (abc.net.au)
AI is increasingly becoming a gateway through which people find information. Audiences still resist having AI produce their news. (niemanlab.org)
Nvidia holds a central position in the AI industry, making disruption to the company potentially catastrophic for the wider ecosystem. The race for advanced AI is also intensifying economic and geopolitical conflict, while concentrated corporate power and infrastructure face growing political resistance. (france24.com)
LLM inference can require reading about 26 GB of model weights to generate each token from a 13-billion-parameter model. The explanation shows how model size drives memory bandwidth demands during text generation. (youtube.com)
Deletion Tuning matches full fine-tuning across three LLM post-training methods by zeroing fewer than 0.05% of model weights. Its compact masks also preserve prior-task accuracy better than full fine-tuning, LoRA, and random sparse tuning. (arxiv.org)
Today's best Hacker News stories, summarized and screenshotted, one email a day.