Meta's Watermelon, the internal codename for its next frontier large language model after Muse Spark, remains unreleased as of early October 2026. Alexandr Wang stated in a July internal town hall that the model, trained with roughly 10x the compute of its predecessor, had reached parity with OpenAI's GPT-5.5 on unnamed benchmarks while still in training. Reports from August cited internal documents targeting an October release, but the model did not appear at Meta Connect in September. Trader sentiment reflects uncertainty over the exact timeline, verification of capabilities, and whether Watermelon will first appear in public text arenas or leaderboards rather than through direct Meta channels. Upcoming catalysts include any pre-training pause signals or official announcements that could shift release expectations.