Meta's Muse Spark 1.1 takes on OpenAI and Anthropic with low-cost coding AI
Meta's new coding model is cheaper than rivals and got Zuckerberg back on X.
Meta publicly launched Muse Spark 1.1 on Thursday, a multimodal AI model designed for agentic coding that directly competes with OpenAI's GPT-5.6 Luna and Anthropic's Claude Haiku 4.5. Priced at $1.25 per million input tokens and $4.25 per million output tokens, it sits slightly above those rivals but offers strong performance in handling complex enterprise tasks like large code migrations, bug fixes, and multi-step reasoning. The model can orchestrate workflows across external apps and services, making it a viable option for businesses automating development pipelines. Meta first announced Spark in April, and this updated version focuses on tool use and computer use capabilities.
CEO Mark Zuckerberg broke a three-year silence on X (formerly Twitter) to promote Muse Spark 1.1, calling it "a strong agentic and coding model at a very low price" and hinting at "more to come soon." The launch comes amid a busy week for AI announcements: Meta also unveiled Muse Image, a new image-generation model; SpaceXAI released a new version of Grok; and OpenAI dropped GPT-5.6. Meta's emphasis on affordability and enterprise-grade agentic features underscores its strategy to undercut competitors while delivering practical automation for developers. The AI coding arms race is heating up, with pricing and capability becoming key differentiators.
- Muse Spark 1.1 costs $1.25/M input tokens and $4.25/M output tokens, slightly above but competitive with OpenAI and Anthropic.
- It specializes in agentic coding, multistep reasoning, tool use, and enterprise workflow automation like code migrations and bug fixing.
- Zuckerberg ended a 3-year hiatus on X to promote the model, signaling its strategic importance for Meta's AI ambitions.
Why It Matters
Enterprises gain a powerful, low-cost coding agent that automates complex workflows and migrations, intensifying the AI pricing war.