Meta's Watermelon model is bigger than its Avocado model.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judge as of 2026-09-15. True: Meta disclosures or reliable reporting show that the model called Watermelon is larger than the model called Avocado, by parameter count or by another size measure the source states (compute, model scale). Mostly true: Watermelon is larger on the main measure, with a small caveat (sizes are only described roughly, or it is larger only in total and not active parameters). Mixed: the evidence conflicts, or the result depends on which measure is used. Mostly false: the two are about the same size, with Watermelon not clearly bigger. False: Watermelon is smaller than Avocado.
Set before the research began.
Evidence (4)
-
E1 Facebook parent Meta has unveiled a new artificial intelligence (AI) model that will compete with Op.. - MK MK (Maeil Business Newspaper) · 13 October 2026
“The Muse Wonder is larger and has a higher operating cost than the existing small model 'Muse Spark'.”
-
E2 Meta Catches Up to OpenAI: Watermelon Model Rivals GPT-5.5, Uses Ten Times More Power | Jarvis AI Jarvis AI · 3 July 2026
“"Watermelon, our next model after Avocado, is currently in training. It uses an order of magnitude more computational power than Avocado,"”
-
E3 Meta Avocado · LLM Releases LLM Releases · 13 March 2026
“Recent reporting identifies the public productized model as Muse Spark, now tracked separately in the catalog.”
-
E4 Research|META: Muse Wonder Returns Meta to the Frontier as Compute Tightens FUNDA · 13 October 2026
“On October 12, Meta Superintelligence Labs (MSL) released Muse Wonder and Muse Wonder Full, the models previously referred to in reports as Watermelon.”
Reasoning
E3 links the Avocado codename to the released Muse Spark model, and E4 says Muse Wonder is the model earlier reported as Watermelon. E1, an independent report, says Muse Wonder is larger than Muse Spark and costs more to run. E2 is a secondary report quoting a statement, from an unnamed speaker, that Watermelon uses an order of magnitude more computational power than Avocado [E2]; compute is a size measure the criteria accept. E1 and E4 were published after the judge date, but they describe the same models, and together the sources consistently show that Watermelon is larger than Avocado.
Cite this claim
Mark Zuckerberg, “Meta's Watermelon model is bigger than its Avocado model.” (said 15 September 2026). Verdict: resolved true. Reliability API, claim c_23c58a58b087, method 0.1.0+3f84d14d, dataset release v1.1.2. https://epistacklabs.com/claims/c_23c58a58b087
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Mark Zuckerberg
-
Mostly true
“We're seeing early progress with the launch of our AI video editing tools across Meta AI and our new Edits app, and there's a lot more to do here.”
Meta launched AI video editing tools across Meta AI and its new Edits app.
-
Mostly true
“This year, Llama 3 is competitive with the most advanced models and leading in some areas.”
In 2024, Llama 3 is competitive with the most advanced models and leading in some areas.
-
Mostly true
“One that people are going to find pretty wild is that it now generates high quality images so quickly that it actually generates it as you're typing and updates it in real time.”
-
Resolved true
“we didn't prioritize coding because people aren't going to ask Meta AI a lot of coding questions in WhatsApp.”
-
Mostly false
“So anytime it is recording, it’s flashing a very visible light.”
Meta's glasses flash a very visible light whenever they are recording.
Verdicts are made automatically; how verdicts are made.