Meta Launches Muse AI Model Update to Target Coding Market

2026-07-10

The BareStory

Meta introduced a significant update to its artificial intelligence model on Thursday, releasing Muse Spark 1.1—referred to in some market reports as Muse Park 1.1. The launch represents Meta's strongest model designed for coding and agentic tasks as the company seeks to compete directly with rivals such as OpenAI and Anthropic.

The company is offering the Muse Spark 1.1 application programming interface (API) through a developer portal under a public preview. While access is currently restricted to Meta's own properties rather than third-party platforms, the shift to a paid proprietary model marks a departure from Meta's previous AI strategy of releasing its Llama family of models to the open-source community. Meta's AI chief, Alexandr Wang, stated that the company remains committed to open source and is developing a variant of Muse Spark to release openly in the future.

According to Wang, the pricing for the new model is positioned to be highly competitive, with input costs set at $1.25 per million tokens and output at $4.25 per million tokens. Wang claimed that Muse Spark 1.1 outperformed rival models in specific tasks involving third-party coding products. He also revealed that Meta's Superintelligence Labs trained the model to boost the capabilities of autonomous AI agents, and that the company is currently training an even more powerful model code-named Watermelon.

Left Perspective

Erode the Open-Access Commons
Discount the Open-Source Promise
Monopolize the Agentic Future

Right Perspective

Incentivize High-Capital Innovation
Calibrate High-Efficiency Pricing
Secure the Strategic Frontier

Left Perspective

• Erode the Open-Access Commons The foundational value of technological democratization is threatened when major tech firms transition from open-source contributions to closed, paid proprietary models. Forcing developers into a restricted public preview on Meta's own properties, rather than utilizing third-party platforms, represents a retreat from collaborative innovation. By locking this advanced coding model behind a paywall, Meta risks centralizing technological progress and limiting the grassroots developer community's ability to audit, modify, and freely build upon foundational code.

• Discount the Open-Source Promise The assertion that a future open-source variant of Muse Spark is still in development functions as a rhetorical shield to pacify the developer community while capitalizing on proprietary monetization. Relying on future promises from leadership while immediately deploying a paid API model creates a two-tiered ecosystem where the public open-source community receives delayed, less capable tools. This strategy prioritizes immediate corporate market capture over the long-term, equitable distribution of AI capabilities.

• Monopolize the Agentic Future Training advanced models specifically to boost the capabilities of autonomous AI agents within a closed ecosystem threatens to create dangerous corporate dependencies. When powerful agentic tools and upcoming models like "Watermelon" are kept proprietary, it concentrates systemic influence over digital workflows in the hands of a single mega-corporation. Without open-source oversight, the deployment of highly autonomous systems raises significant concerns regarding accountability, bias, and the monopolization of next-generation digital labor.

Right Perspective

• Incentivize High-Capital Innovation Systemic stability and technological leadership require immense capital investment, which can only be sustained through viable monetization strategies. Transitioning to a paid proprietary model like Muse Spark 1.1 allows Meta to recoup the massive R&D costs incurred by specialized divisions like Superintelligence Labs. Securing a direct return on investment is the most reliable engine for funding next-generation computational infrastructure and pushing the boundaries of AI capabilities.

• Calibrate High-Efficiency Pricing Undercutting established competitors like OpenAI and Anthropic with highly competitive pricing—specifically $1.25 per million input tokens and $4.25 per million output tokens—drives market efficiency and benefits the broader economy. Outperforming rivals in specific third-party coding tasks proves that disciplined, market-oriented development yields superior, cost-effective products. This aggressive pricing strategy forces industry-wide cost discipline, making cutting-edge coding tools highly accessible to commercial enterprises.

• Secure the Strategic Frontier Maintaining tight proprietary control during the critical development phase of autonomous agents and powerful successor models like "Watermelon" is a strategic necessity for national and corporate security. Prematurely releasing highly capable coding and agentic models into the open-source domain risks intellectual property theft and the rapid proliferation of powerful tools to hostile actors without guardrails. A controlled, API-restricted rollout ensures that advanced technology is safely stabilized, stress-tested, and commercially secured before wider deployment.

How it may affect me

As a U.S. reader:

• You may find access to advanced AI development tools temporarily restricted to Meta-owned properties, limiting your ability to utilize this technology on third-party platforms during the public preview phase.

• In the short term, if you are a developer or business owner, you could benefit from lower operational expenses due to Meta's competitive pricing of $1.25 per million input tokens and $4.25 per million output tokens compared to rival models.

• In the long term, you may experience a shift in the tech landscape as major companies move away from open-source models toward paid, proprietary systems, potentially reducing the availability of free, collaborative software tools.

• You could eventually interact with more advanced, autonomous digital agents and highly capable automated workflows as Meta and its competitors use proprietary profits to fund and deploy successor models like Watermelon.

Meta Launches Muse AI Model Update to Target Coding Market

The BareStory

Left Perspective

Right Perspective

How it may affect me

Read the story at