Meta AI Launches Muse Ember: A Lightweight, Efficient Model for Agents
Meta AI has introduced Muse Ember, its smallest and most efficient model designed specifically for agentic execution tasks such as tool calls, coding subtasks, and long document parsing. It features a 1-million-token context window and is priced competitively alongside similar lightweight models like GPT-6 Luna and Claude Haiku 5.5. Muse Ember provides developers with a high-performance, cost-effective option for running repetitive AI agent workflows at scale. Its release intensifies competition in the lightweight model category, offering strong intelligence benchmarks relative to its operational cost. For prompts up to 200k tokens, Muse Ember 1.0 costs $0.10 per million input tokens and $0.50 per million output tokens, but prices increase to $0.30 and $1.50 respectively for prompts exceeding 200k tokens. Benchmarks from Artificial Analysis show that it achieves high intelligence index scores relative to its overall cost per task.
## BACKGROUND
AI agents are autonomous systems that leverage large language models to chain together complex operations, interact with external software through tool calls, and parse extensive documents. Context windows measure the maximum amount of text a model can process at once, while input and output token pricing determines the operational cost of utilizing these models via an API.