Been digging into the numbers on this one. Muse Spark 1.1 (Meta's new model, first one they've ever actually charged for API access) is priced at $1.25/$4.25 per million tokens. Compare that to Opus 4.8 ($5/$25) or GPT-5.5 ($5/$30) and it's not close, 4x cheaper on input, 6-7x on output. Genuinely useful if you're building something that doesn't need the absolute smartest model for every call. Not everything needs the frontier flagship, cheap-but-good-enough is a real category and Meta wasn't really in it before this.
Except that comparison is entirely API pricing. Most people don't touch the API, they're on a Claude or ChatGPT subscription. And when I went looking for the Muse Spark equivalent of Claude Pro or ChatGPT Plus, there isn't one. Meta's only consumer option is free chat through the Meta AI app, rate limited, no paid tier at all. So the "we're way cheaper" pitch only actually applies if you're a developer paying per token, which is a much smaller group than the framing implies. Kind of feels like they picked the one lane where they can currently win and built the whole announcement around it.
Also can't fully separate this from what happened two days before Muse Spark launched. Meta's other new model, Muse Image, shipped a feature that let anyone tag a public Instagram account and pull that person's photos straight into an AI generation, opt out instead of opt in, no notification. SAG-AFTRA called it out, Meta pulled just that one feature within days. Same lab, same week. Doesn't cancel out the pricing thing being genuinely good, but it's hard to get excited about "cheapest agentic model" news from a team that shipped that in the same breath.
Not sure where that leaves the overall verdict. Price and raw capability both look fine on their own. It's the packaging around both of them, what got compared to what, what got shipped without a second thought, that I keep getting stuck on.
[link] [comments]