Meta is preparing a significant upgrade to its Muse Spark artificial intelligence model, with Chief AI Officer Alexandr Wang confirming that the next iteration will feature substantial improvements in coding and agentic capabilities. The update, internally codenamed Watermelon, represents a strategic push by Meta to close the performance gap with rival platforms like OpenAI and expand its footprint in the enterprise AI market.
Catching up to GPT 5.5
According to reports citing anonymous sources, the Watermelon model utilizes significantly more computing power than its predecessor. Wang stated that this increased compute has allowed the model to catch up with OpenAI’s flagship GPT 5.5. This development comes shortly after CEO Mark Zuckerberg addressed concerns about the pace of AI agent development during a company townhall, prompting Wang to clarify the progress via social media.
The update is expected to roll out soon through Meta AI and a new API interface. By enhancing these specific areas, Meta aims to offer a competitive alternative for developers and businesses currently relying on other leading models.
Implications for enterprise adoption
For enterprises, the arrival of a stronger Meta model could reshape the AI landscape. Pareekh Jain, principal analyst at Pareekh Consulting, noted that increased competition often leads to lower AI costs and provides organizations with viable alternatives to OpenAI and Anthropic. If Meta offers Watermelon as an open-weight or low-cost option, it could make AI coding assistants more affordable while giving companies better control over their data and reducing vendor lock-in.
This timing aligns with broader industry pressures. GPU shortages, high licensing fees, and rising inference costs have made access to top-tier coding models increasingly expensive for many businesses. A competitive model from Meta could alleviate some of these financial burdens.
Shifting toward AI-native platforms
Analysts suggest this update signals Meta’s intent to move beyond providing just foundation models. Charlie Dai, principal analyst at Forrester, indicated that Meta appears interested in becoming a platform for building AI-native applications and agents. This strategy is supported by recent acquisitions, including efforts to acquire Manus, and initiatives like Pocket, which aim to lower barriers for creating software with less technical expertise.
Meta is also reportedly developing new cloud infrastructure business lines that would sell access to AI computing power and models. This broader push into the enterprise market suggests a long-term strategy to enable business users to build workflow automations and lightweight applications more easily.
Execution challenges remain
Despite the potential benefits, analysts caution that enterprise adoption will not be automatic. Dai emphasized that Meta must prove superior real-world coding quality, reliable agent execution, and strong security governance to win over IT professionals. Additionally, geopolitical and regulatory considerations outside North America are increasingly influencing model choices.
To succeed, Meta needs to deliver compelling customer outcomes and establish strong local partnerships. The upcoming rollout of Watermelon will be a critical test of whether the company can translate its technical improvements into tangible value for developers and enterprises.
Source: Computerworld
Over to you: Would you trust Meta’s new Muse Spark model for critical coding tasks over established options like OpenAI?
You may also like
