Meta Superintelligence Lab released Muse Ember on Monday at $0.10 per million input tokens, and in doing so landed on precisely the number two rivals already charge. Claude Haiku 5.5 is $0.10 per million input for prompts under 100,000 tokens. OpenAI’s gpt-6-luna is $0.10. Three labs, three budget models, one price.
Ember is the fourth model in the Muse line, after Spark in April, the open-source Glimmer in August and Wonder in October. Meta positions it as the execution half of a pair: Wonder plans, Ember handles the volume. It takes text, images and audio, and Meta cites the Artificial Analysis Intelligence Index to argue it is both cheaper and smarter than the budget models it is priced against.
What the convergence means practically is that price has stopped being the thing that picks a cheap model. At $0.10 everywhere, the decision moves to what you actually need: Haiku 5.5 jumps to $0.50 input once a prompt passes 100,000 tokens, which matters if you are feeding it long documents. Ember handles audio, which Haiku does not. Luna sits inside the ChatGPT free tier, which matters if your users are already there.