Meta launched Muse Spark 1.1 on July 9, 2026, a multimodal AI model designed for agentic coding and programming tasks, available to US developers through a public Meta Model API preview [1, 2, 3]. The model offers advanced multi-step reasoning, bug fixing, digital workflow management, and supports multi-agent systems, handling multiple data types including images and videos [1, 2, 4].
This marks Meta's first AI model offered as a paid API service targeting enterprise developers, shifting from open-source distribution to monetizing AI via API usage fees [5, 6, 3]. Meta CEO Mark Zuckerberg announced the launch in his first X (formerly Twitter) post in three years, emphasizing Muse Spark 1.1's strong agentic, tool-use, and coding performance at a low price [2, 7, 4]. Zuckerberg said, "Muse Spark 1.1 delivers exceptional performance in personal agentic tasks that require planning and orchestration across a range of external apps and services" [2].
Meta is pricing the API at $1.25 per million input tokens and $4.25 per million output tokens, with $20 in free credits for new accounts [2, 6, 3]. This price is roughly 25% of top competitor AI model costs, aiming to attract enterprise users concerned about rising AI usage expenses [6, 3, 8]. Zuckerberg noted, "Since this is not an open-source model, this is, I think, the first time that we're doing a real serious API. And the pricing is going to be very aggressive and attractive" [3].
The Muse Spark 1.1 model features a very large context window of up to one million tokens, boosting its ability to handle long conversations, documents, and complex autonomous tasks [4]. It reportedly matches or outperforms OpenAI's GPT-5.5 and Anthropic's Claude Opus 4.8 on coding and agentic benchmarks, though some sources consider Meta still behind these rivals [4, 8, 2]. Meta’s AI lab head Alexandr Wang called it "the lab’s strongest model for agentic and coding work yet" [3].
Meta positions Muse Spark 1.1 and the new Meta Model API as managed cloud-hosted inference services, unlike prior models that were available for self-hosting [3]. Internally, Meta employees are already using Muse Spark 1.1 to develop applications and features, demonstrating its practical utility [6].
This launch follows Meta’s July 7 release of Muse Image, an AI image generation model integrated into Instagram, WhatsApp, and the Meta AI app [1, 2, 6]. Meta has increased its 2026 AI capital expenditure guidance to $125-$145 billion, indicating significant investment in AI research and expecting AI monetization to grow as a revenue source [7].
Muse Spark 1.1 is currently available only to US developers, with no announced timeline for global availability [1, 2, 3].
