Sources: Meta has repeatedly delayed Muse Spark API's release from April due to bugs and infrastructure needs; Meta now says it plans to release the API in June
The shifting timeline is a setback for Meta's ability to monetize its massive investments in building frontier AI models
Wall Street JournalMeghan Bobrowsky
Context & Ripple Effects
Muse Spark was introduced as a Meta Superintelligence Labs model already used in Meta AI and its shopping mode, with a private API preview for selected partners preceding a planned broader paid offering. Meta also said it intended to open-source a version of the model.
The delayed API turns a product-distribution plan into an execution test: Meta can use the model inside its own products, but external monetization depends on stable infrastructure and a reliable developer service.
First-order effects
Meta’s wider paid API launch moves from the earlier April target to June, delaying access for developers and partners beyond the private preview.
Bugs and infrastructure requirements force Meta to prioritize operational readiness before it can convert Muse Spark’s internal deployment into API revenue.
Second-order effects
Partners evaluating Muse Spark for production use may defer integration timelines or keep alternative model providers in their stack until Meta demonstrates dependable API availability.
The delay raises the importance of Meta’s infrastructure build-out: capacity and service reliability become constraints on monetization, not just model quality.
Third-order effects
If recurring, such delays would underline that frontier-model competition is increasingly decided by the ability to operate developer-facing services at scale, alongside training and deploying models in consumer products.
Meta’s simultaneous plans for paid access and an open-source version illustrate a dual-distribution strategy whose commercial effectiveness will depend on whether proprietary hosted access offers sufficiently reliable, differentiated capabilities.
The trend: Frontier AI builders are moving from model announcements and internal deployment toward the harder business of operating reliable, monetizable APIs for outside developers.
Scoop: Meta has delayed the release of its Muse Spark API multiple times, according to ppl familiar with the matter. As of Tues, Meta didn't have a set release date for it. On Weds in response to my qs, Meta said it plans to release the API this month without providing details. […
Meta can't even provide working API access to their new model. You dolts want to spend $125 billion or more on capex this year and you can't work out the basics? Somebody take Mark's wallet. Enough! [embedded post]
In April, Alexandr Wang posted that the API was “coming soon.” https://x.com/... Since then, it has been delayed multiple times due to bugs that became apparent in testing and the need to build more infrastructure, ppl familiar with the matter said.
The API would provide another revenue stream for Meta, which plans to spend up to $145 billion on capex this year. Read more in my story here: https://www.wsj.com/...
Hatch is also being used to refer to the weird tamagotchi pendant thing they're building, which will also connect to your email and calendar. From everything I've heard this is a product for aliens
I'm so tired of the $META “we're getting into B2B take us seriously this time” trope? Every fiber of their soul is consumer ads. Meta's list of B2B/commercial fails: 1.Workplace from Meta (2016-2026): Enterprise collaboration tool meant to take on Slack and Teams. Landed
SCOOP: Meta is considering charging up to $200 a month for its planned consumer version of the OpenClaw AI agent tool, according to internal documents. Hatch will be able to vibe-code new software tools. Users can describe what they want in plain language, such as “build me a
Meta has repeatedly delayed broad developer access to Muse Spark, its first closed AI model, though the company says it is testing the API with early partners and expects to release it this month. …