Meta's 'Watermelon' matches GPT-5.5 performance in internal tests
Meta's upcoming AI model, internally codenamed 'Watermelon', reportedly matches the performance of OpenAI's GPT-5.5 on internal benchmarks.
Vad har hänt
According to The Economic Times, citing Business Insider, Meta's upcoming AI model, 'Watermelon', has achieved performance levels comparable to OpenAI's GPT-5.5 in internal benchmarks. Meta's AI lead, Alexandr Wang, informed employees of this during an internal meeting. The model is still in the training phase and is intended to succeed Meta's previous frontier AI model, 'Avocado'.
Key facts
| Modellnamn | Watermelon |
|---|---|
| Jämförbar prestanda med | OpenAI GPT-5.5 |
| Datum för rapport | 6 juli 2026 |
| Föregångare | Avocado (Muse Spark) |
| Beräkningskraft | 10x mer än Avocado |
”Meta's upcoming artificial intelligence (AI) model, codenamed 'Watermelon', has reached performance comparable to OpenAI's GPT-5.5 on internal benchmarks, Meta's chief AI officer Alexandr Wang told employees during an internal town hall, Business Insider reported, citing sources.”
”Wang reportedly said the model caught up with GPT-5.5 on key benchmarks, but did not specify which benchmarks were used or disclose evaluation data.”
”The report added that Watermelon is being trained using around ten times more computing power than Avocado.”
Varför det spelar roll
This development signals Meta's intensified position in the AI race and its ambition to compete directly with leading players such as OpenAI. The performance comparison with GPT-5.5, despite the lack of specified benchmarks, suggests significant progress. This could strengthen Meta's product offering and innovation within generative AI.
Vem påverkas
Meta and OpenAI are the primary parties directly affected by this news, as it highlights their competition in AI development. Developers and researchers in the AI field are also closely monitoring such advancements. Users may be indirectly affected through improved AI-driven features in Meta's products, such as Muse Spark.
EU-status
Not applicable for EU status. The news concerns internal development and performance benchmarks rather than specific launches within the EU.
Mer att veta
Alexandr Wang did not specify which benchmarks were used to assess 'Watermelon's' performance against GPT-5.5. The model is being trained with approximately ten times more computing power than 'Avocado', which was publicly launched as Muse Spark.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilka bolag berörs?
Länken öppnar i nytt fönster och leder till utgivarens egen sida.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.