Meta has delayed the release of its proprietary Muse Spark AI API, slowing a key move away from open-source AI and raising fresh questions about monetising its massive AI investments.
Meta has repeatedly delayed the release of the Muse Spark AI model API, creating a setback for its planned shift away from open-source AI and potentially slowing the monetisation of its growing artificial intelligence investments.
The API was originally expected to launch alongside the model in April 2026 but has been postponed multiple times. According to reports, testing bugs and infrastructure requirements have contributed to the delays, leaving developers unable to build commercial applications on the model through an official API.
The delay is significant because Muse Spark represents a strategic departure from Meta’s traditional open-source AI approach. The proprietary model reportedly “represents a shift from open-source software”, signalling a move towards a business model more closely aligned with rivals OpenAI and Anthropic.
The postponement is also raising concerns about Meta’s ability to generate returns from its projected US$145 billion capital expenditure on AI infrastructure. Delays to the API rollout could push back revenue opportunities tied to the model and increase investor scrutiny of the company’s AI spending plans.
Despite the setback, sources cited in the report indicate that Meta still expects to launch the API in June 2026, suggesting the delay is temporary rather than a cancellation.
Internal benchmark evaluations reportedly position Muse Spark against leading models from OpenAI and Anthropic, underscoring Meta’s ambitions in the premium AI market.
Alexandr Wang, Chief AI Officer at Meta, had previously said the API was “coming soon”.
