Unverified50% confidenceFactExact time
Meta reported experimental results stating that Wonder Full's performance on unfamiliar tasks improved relative to the corresponding trend of the previous-generation model, which Meta attributes to a stronger base model.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedMeta此次同时推出两个版本，命名逻辑暗示了能力分层。'Full'通常代表完整参数版本，面向研究或高算力场景；而标准版Muse Wonder更可能针对实际部署做了优化，在延迟、成本和吞吐之间取得平衡。72% similarUnverifiedMeta的LLaMA和微软的Phi系列开源小模型证明精心设计的小模型在特定任务上可接近大模型表现70% similarUnverifiedThe performance of lightweight models on specific tasks is rapidly approaching or even surpassing the previous generation of midweight models.68% similarUnverifiedMeta's Llama series of models publicly acknowledged using some data generated by stronger models in their training.67% similarUnverifiedMeta and associated labs have introduced the Dream model, which completes image understanding and text-to-image generation with a single model by first generating a draft and then self-evaluating and refining it, with processing speed reportedly improved by approximately 10%.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1067544API
curl https://kongchang.com/api/v1/knowledge/claims/1067544MCP
get_claim(id=1067544)