Sources: Alexandr Wang said Meta's model currently in training, codenamed Watermelon, matches GPT-5.5 and uses an "order of magnitude more compute than Avocado" (Business Insider)
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Alexandr Wang, Meta's superintelligence chief, told employees the company's model currently in training, codenamed 'Watermelon,' matches the capabilities of GPT-5.5.
- Wang said 'Watermelon' uses an 'order of magnitude more compute' than 'Avocado' and framed Meta as 'making significant progress in the AI model race.'
Why it matters: This is an internal pep-talk claim from Meta's superintelligence chief, not an independently verified benchmark result. The 'order of magnitude more compute' disclosure signals Meta is leaning heavily on raw compute spending to close the gap with frontier models, yet the truncated source offers no release timeline or external validation.
