Nội dung
Meta’s announcement that its new AI model has reached an important OpenAI milestone is more than just a technology story. For Vietnamese marketers, it signals a rapidly accelerating race in infrastructure, generative content capabilities, and ad automation, bringing changes to how platforms optimize campaign performance. As the AI giants draw closer in capability, competitive advantage will shift more toward data, distribution, and real-world application.
-
Key points:
- Alexandr Wang said Meta’s internal Watermelon model has caught up with OpenAI’s GPT-5.5 on some closely watched benchmarks.
- Meta is accelerating investment in compute, AI talent, and data center infrastructure to close the gap with OpenAI, Google and Anthropic.
- Wang said an update to Muse Spark will launch soon, with a focus on coding and agentic capabilities.
- For the advertising industry, this signals that Meta could continue improving AI tools for creative support, optimization, and campaign automation.
Meta signals a major step forward in the AI model race
In an all-hands internal meeting, Alexandr Wang — who leads Meta’s superintelligence division — said the upcoming model called Watermelon has caught up with OpenAI’s GPT-5.5. According to a source cited by Business Insider, this assessment was based on closely tracked AI benchmarks, though it is not clear which specific benchmark Wang was referring to. This is a notable milestone because Meta has long been seen as trailing the leaders in generative AI.
Wang also revealed that Watermelon is still in training and is using significantly more compute than Avocado, the internal name for the previous model line. In other words, Meta is not just making small tweaks to its models — it is dramatically increasing resource investment for the next generation.
Why this message matters for the advertising industry
For the advertising market, any improvement in Meta’s foundation models could flow into the product ecosystem businesses use every day: from helping write content and generate ad variations to understanding user intent and optimizing automated tasks. As Meta closes the gap with leading AI rivals, the company’s ability to integrate these capabilities more deeply into digital advertising is significant.
It is important to note that this is not yet confirmation from OpenAI or an independent technical announcement. Still, Wang’s remarks suggest Meta wants to send a clear signal to the market: its major investments in chips, data centers and specialized talent are beginning to produce concrete results.
Meta’s strategy: infrastructure, talent and pressure to prove capability
Meta has long pursued the goal of narrowing the gap with OpenAI, Google and Anthropic. The company has spent heavily on infrastructure and teams, while betting on building an elite AI group inside Meta Superintelligence Labs. Wang currently leads the TBD research group and several other AI areas, underscoring his central role in this strategy.
According to the source material, Meta is also willing to spend hundreds of millions of USD to attract top AI talent. At the same time, the company has raised its capital expenditure forecast for this year to 125–145 billion USD for chips, data centers and related infrastructure. In a race where new model launches are measured in months, the pressure on Meta is not only to catch up, but also to prove that massive investment can be turned into useful products for users and businesses.
Wang hints at upcoming upgrades for Muse Spark
On X, Wang also hinted that an update for Muse Spark will arrive soon, with significant improvements in coding and agentic capabilities. As a model becomes better at programming and executing goal-oriented tasks, its applications typically expand not only in developer tools but also in marketing automation systems, landing page creation, content generation and campaign assistants.
He also told users that Meta will have a coding model on par with Anthropic’s Claude Opus “pretty soon.” While this is more of a directional statement than a technical commitment, it shows Meta wants the market to see it as a serious contender in the premium model segment, rather than just a social platform with AI features.
What this means for the Vietnamese market
For Vietnamese businesses and marketers, this development has three practical implications. First, the cost of experimenting with AI in advertising could continue to fall as major platforms push harder on content generation and campaign optimization features. Second, competitive advantage will no longer lie in whether a company “uses AI,” but in how it organizes data, creativity and workflows so AI can deliver real results. Third, marketing teams should closely follow Meta’s updates, since this is the platform with the most direct impact on many paid social campaigns in Vietnam.
In the short term, it is still too early to conclude that Watermelon or Muse Spark will change the game immediately. But if Meta has truly caught up in a meaningful part of OpenAI’s capabilities, the digital advertising market is likely to enter a deeper phase of AI adoption, where deployment speed and content personalization become decisive factors.
Source: Business Insider.
See more marketing news and guides at https://marketing365.vn.
Follow more updates from Marketing365 to stay up to date with the latest marketing trends.
Read more articles in the same category Digital Trends.
This article focuses on Meta catches up with OpenAI with a perspective for the Vietnamese market.
