Artificial Analysis data backs Meta's pitch: at $0.10 per million input tokens, Ember is presented as cheaper and smarter than Haiku and Luna on cost per task, with Wonder planning and Ember handling the volume.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [x.com](https://x.com/finkd/status/2119851180948152590)
Why it matters #
Agent workloads turn model choice into a volume business: thousands of small calls for routing, extraction and tool use can cost more than the one hard reasoning step. Ember now has a concrete price, $0.10 per million input tokens against Spark at $1.25, and Artificial Analysis cost-per-task data behind Meta's claim that it beats Haiku and Luna. That gives Meta a cheap default inside Muse Code and on its API for the work agents repeat all day.
Mark Zuckerberg (@finkd) released Muse Ember on November 9th, announcing in a post on X that Meta's smallest Muse model is available on the Meta Model API and in Muse Code.
…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.