Mark Zuckerberg returns to X after 3 years to unveil Meta’s Muse Spark 1.1 AI model, developer API

Meta CEO Mark Zuckerberg returned to X (formerly Twitter) for the first time in more than three years on Thursday to announce the launch of Muse Spark 1.1, the company’s latest artificial intelligence model, alongside the public preview of the new Meta Model API for developers.

The posts marked Zuckerberg’s first activity on the platform in over three years and signaled Meta’s latest push to compete directly with OpenAI and Anthropic in the rapidly expanding market for enterprise AI models.

“Today we’re releasing Muse Spark 1.1 — a strong agentic and coding model at a very low price. It’s available through our new Meta Model API and in Meta AI,” Zuckerberg wrote.

In another post, he highlighted the model’s capabilities, saying, “Muse Spark 1.1 is strongest at agentic performance, tool use, and computer use. It does well on long-running tasks with 1M token context window, can delegate execution to sub-agents running in parallel, and is trained to use computer interfaces on desktop, mobile, or browser.”

He added that the Meta Model API would allow developers to build applications using Muse Spark for the first time.

“The Meta Model API allows developers to build using Muse Spark for the first time. Our focus is on delivering strong agentic and multimodal models at very low cost. More to come soon,” Zuckerberg posted.

Meta opens developer access

Separately, Meta announced that developers in the United States can now access Muse Spark 1.1 through a public preview of the Meta Model API.

The company described the upgraded model as its most capable AI system yet for coding, agentic workflows and software automation. According to Meta, Muse Spark 1.1 can write and debug code, interact with software tools, understand text, images and videos, and complete complex multi-step tasks with minimal human intervention.

The launch follows Meta’s debut of the original Muse Spark model in April, developed by its Superintelligence Labs as part of the company’s effort to narrow the gap with AI rivals.

Meta takes on OpenAI and Anthropic

The release also marks Meta’s formal entry into the paid AI model API market, placing it in direct competition with OpenAI and Anthropic.

Developers signing up for the Meta Model API reportedly receive $20 in free credits before transitioning to pay-as-you-go pricing.

Meta has reportedly priced the service at $1.25 per million input tokens and $4.25 per million output tokens. The pricing is higher than OpenAI’s entry-level GPT-5 mini and Anthropic’s Claude Haiku 4.5, but below Anthropic’s premium Claude Sonnet 4.6 model.

Expanding Meta AI ecosystem

Muse Spark 1.1 is also being rolled out in Thinking mode within the Meta AI app and website. The model is expected to gradually replace existing Llama models powering AI assistants across WhatsApp, Instagram, Facebook, and Meta’s AI-enabled smart glasses.

The launch comes just days after Meta expanded its generative AI portfolio by introducing Muse Image, its first image-generation model developed by Meta Superintelligence Labs, as the company accelerates its strategy to build what it describes as “personal superintelligence.”