Event ·Multi-source

Meta发布新模型，在多项基准测试中超越GPT-4

Meta公司发布了名为Llama 3.2的新模型，在MMLU和HellaSwag等基准测试中取得了比GPT-4更好的成绩。该模型是针对企业用户设计的，并提供了免费和付费两种版本。Meta表示，新模型在推理任务上表现更优。

Current read

Meta搞了个新模型，叫Llama 3.2，在MMLU和HellaSwag这些测试里比GPT-4强，还分了免费和付费版，企业用户可以试试。

11 sources37° AI heatLast updated 2026/9/18 14:31:21

Evidence

11 sources

Primary sourceThe Rundown AI

2026/9/18 14:31:21View source

Related source 1AI Engineer

2026/9/16 19:29:32View source

Related source 2宝玉

2026/9/18 04:59:54View source

Related source 3小互

2026/9/17 02:10:49View source

Related source 4掘金本周最热

2026/9/17 04:51:39View source

Related source 5theverge

2026/9/17 19:28:24View source

Related source 6OpenRouter

2026/9/17 23:24:41View source

Related source 7arXiv cs.LG

2026/9/18 12:39:47View source

Related source 8Gary Marcus

2026/9/18 21:58:14View source

Related source 9量子位

2026/9/19 11:43:12View source

Related source 10Nathan Lambert: Interconnects

2026/9/19 15:42:20View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story