Meta is making significant progress in the AI model race, its superintelligence chief Alexandr Wang told employees today. According to sources cited by Business Insider, Wang stated that the company's model currently in training, codenamed Watermelon, matches the capabilities of GPT-5.5.
This development signals Meta's aggressive push to compete with OpenAI at the frontier of artificial intelligence. The Watermelon model is reportedly using an “order of magnitude more compute than Avocado,” suggesting a massive scale-up in training resources.
The compute figure highlights the enormous computational investment Meta is willing to make to close the gap with rivals. No specific performance benchmarks or release timeline for Watermelon have been disclosed.
If Wang's claim holds true, Meta could challenge OpenAI's dominance in generative AI. However, internal performance claims often differ from real-world evaluations, and the model's ultimate success depends on future testing and deployment.
Industry analysts caution that unverified claims of parity with GPT-5.5 should be taken skeptically until independent benchmarks confirm the results.