Stock Markets July 2, 2026 08:31 PM

Meta’s Watermelon Model Said to Match OpenAI’s GPT-5.5, Company Officials Say

Alexandr Wang asserts upcoming AI model has closed the gap with competitors; details on benchmarks remain unclear

By Sofia Navarro
Share
Twitter Reddit Facebook LinkedIn
META

Meta Platforms’ lead of its superintelligence team, Alexandr Wang, told employees at a town hall that the company’s forthcoming AI model, codenamed Watermelon, has reached parity with OpenAI’s GPT-5.5 according to individuals familiar with the matter. The company’s progress comes amid a broader push to scale its AI stack after earlier models lagged peers.

Meta’s Watermelon Model Said to Match OpenAI’s GPT-5.5, Company Officials Say
META
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Alexandr Wang said at a town hall that Meta’s Watermelon model has caught up with OpenAI’s GPT-5.5, according to two people familiar with the matter - benchmarks were not specified.
  • Meta has invested heavily - spending billions on chips, data centers, and talent - but has generally trailed AI offerings from OpenAI and Anthropic.
  • Meta released the Muse Spark model in April, which showed strong benchmark results in some areas but was still seen as behind rival models; Mark Zuckerberg is personally overseeing AI development and hired Wang last year to lead the division.

Meta Platforms is reporting internal progress on a next-generation AI model known as Watermelon, with superintelligence chief Alexandr Wang stating at a town hall that the new model has caught up to OpenAI’s GPT-5.5, according to two people familiar with the situation.

Wang’s comment, made to employees during the meeting, described Watermelon as now comparable to a leading rival model. It is not clear which benchmarks Wang referenced when making the comparison, and the individuals who described the town hall did not specify performance metrics.

Meta’s efforts to close the gap follow a period in which the company has trailed other AI providers. Despite substantial investments - including spending billions of dollars on chips, data centers, and talent acquisition - the company’s public AI releases have not consistently matched competing offerings from firms such as OpenAI and Anthropic.

In April, Meta made available its Muse Spark model. Observers noted Muse Spark performed well on certain benchmark tests, but it was still generally regarded as falling short of the leading models in the field.

Leadership at Meta has prioritized accelerating AI development. CEO Mark Zuckerberg has been personally involved in overseeing work on the company’s models, and last year he recruited Wang to run Meta’s AI division.

The company’s internal assessment that Watermelon has reached parity with GPT-5.5 signals a potentially significant step in Meta’s AI roadmap, but the absence of benchmark details leaves the claim without public verification. Until Meta or independent evaluators disclose specific metrics, the comparison cannot be independently confirmed.


Summary

Alexandr Wang told employees that Meta’s upcoming Watermelon model has matched OpenAI’s GPT-5.5, according to two people familiar with the town hall. The exact performance measures Wang used were not disclosed. Meta has previously lagged other AI providers despite major investments and released Muse Spark in April, which performed well on some benchmarks but remained behind peers. CEO Mark Zuckerberg is directly overseeing AI development and hired Wang last year to lead the division.

Risks

  • Lack of disclosed benchmarks - without public metrics, claims that Watermelon matches GPT-5.5 cannot be independently verified; this creates uncertainty for stakeholders tracking AI competitiveness.
  • Prior underperformance - Meta has previously trailed other AI providers despite heavy spending, indicating execution and competitive risks for its AI initiatives.

More from Stock Markets

Monster's New Product Wave Strengthens Near-Term Outlook, Morgan Stanley Says Jul 2, 2026 Zuckerberg Says Progress on Meta’s AI Agents Has Lagged; Company Reassesses Plans Jul 2, 2026 QTS Pulls the Plug on Digital Gateway Data Center in Virginia Jul 2, 2026 Texas man charged with manslaughter after Tesla crashes into Katy home, killing 76-year-old woman Jul 2, 2026 Kioxia to Begin Mass Production of 10th-Gen NAND as AI Demand Drives Rapid Rebound Jul 2, 2026