Meta AI has launched Muse Ember, a new model positioned for complex reasoning and agent work like tool calls, coding subtasks, routing, extraction, and long document parsing.
The service matches the per-token pricing of GPT-6 Luna and Claude Haiku 5.5 for shorter prompts. Both competitors charge $0.10 per million input tokens and $0.50 per million output tokens. Today’s other major release, Claude Fable 5.2, costs 80 times that amount at $8 and $40 respectively.
Pricing becomes less straightforward as prompts extend in length. Luna, Ember, and Haiku all support up to a million input tokens, yet each has a threshold where the price increases. When accounting for that threshold, Ember sits between the other two models.
Cost comparison
Comparing price per token is less useful now because reasoning tokens mean different models might use vastly different numbers of tokens to solve the same problem.
Artificial Analysis offers a more practical measure. Their Intelligence Index vs Cost per Task chart shows the following for those three models:
- Claude Haiku 5.5 (low) scores about 19 at roughly $0.044.
- GPT-6 Luna (non-reasoning) scores about 9.5 at roughly $0.07.
- GPT-6 Luna (xhigh) scores about 23.5 at roughly $0.10.
- Muse Ember (xhigh) scores about 34 at roughly $0.12.
- Muse Ember (max) scores about 36 at roughly $0.14, both on the Pareto line.
- Claude Haiku 5.5 (xhigh) scores about 34.5 at roughly $0.25, well to the right of the Pareto line.
The Muse Ember models sit right on that Pareto line, the point where you can only improve one dimension by giving up another. Meta call this out in their announcement:
It also sits on the efficiency frontier of the full Artificial Analysis leaderboard: no model scores higher for less per task.
That grey dot on the Pareto line in the middle of the green area is GPT-6.1 Terra (low).
Muse Ember has six reasoning levels: minimal, low, medium, high, xhigh, and max. I rendered six pelicans riding bicycles to test them. The “high” reasoning level failed to return a full SVG, and the other SVGs were surprising in that the reasoning level did not cleanly map to the visual quality of the response.
Here is the “max” one, but the others are worth a look too:
The Ember pelicans are comparable in quality to those I got from Muse Wonder, which is 40 times the price of Muse Ember.
