Meta Unveils Muse Ember: A New Budget-Friendly AI Model
Designed for routine tasks, the new model offers high efficiency at a lower cost.
Efficiency and Pricing
Meta has announced a new addition to its artificial intelligence lineup: Muse Ember. According to Mark Zuckerberg, this compact model is optimized for the daily tasks performed by AI agents, including data processing, writing code snippets, document analysis, and interacting with various software tools.
The primary advantage of Muse Ember is its high efficiency. The model is priced at $0.10 per million input tokens and $0.50 per million output tokens. While these prices are comparable to competitors like Anthropic’s Claude Haiku 3.5 and OpenAI’s GPT-4o mini, Meta claims that Ember delivers significantly more work per dollar. Data from the independent research firm Artificial Analysis shows that Muse Ember achieved an intelligence score of 36, matching Claude Haiku 3.5, while offering a cost-per-task that is three times lower.
Technical Capabilities and Safety
Muse Ember is a multimodal model, capable of processing text, images, audio, and video. It can handle up to one million tokens simultaneously, which is sufficient for analyzing extensive codebases. Meta notes that Ember is designed to work alongside their larger model, Muse Wonder: Wonder handles strategic planning, while Ember executes high-volume, routine tasks.
Safety was a priority during development. The model underwent rigorous testing for chemical and biological threats, cybersecurity risks, and potential loss of control. Meta states that in all evaluations, Ember remained well below the high-risk threshold.
Availability
Muse Ember is now available through the Meta Model API and Muse Code platforms. Notably, Meta has chosen not to release the model weights publicly, keeping them closed as with other models in the Muse series. Although it is currently one of the most efficient models on the market, the company confirmed it is already working on more sophisticated and powerful versions. For developers working on integrating AI agents, the availability of cost-effective tools like Ember significantly reduces the expenses associated with automation processes.