Anthropic Mythos-class model checkpoint reportedly leaked on Hugging Face
What happened
On November 1 at 23:45, a Hacker News post said an Anthropic Mythos-class model had reportedly leaked on Hugging Face, and the links were removed over malware risk. The post had 14 points and 4 comments, with no further detail. A follow-up at 00:43 on November 2 said the leaked material was a pretraining intermediate checkpoint of numbat, an Anthropic next-generation Mythos-class training run: about 14T parameters, 114 layers, 772 shards, a hybrid GDN/MLA and MoE architecture with 1,224 experts, top-48 plus one shared expert, in NVFP4 format. The follow-up still says "reportedly" and offers no evidence the leak is confirmed.
Written by AI from the coverage · updated 2 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageAlleged Mythos Class Anthropic Checkpoint Leak
一个据称是 Anthropic 下一代 Mythos 级训练任务 numbat 的预训练中间检查点被泄露，规模约 14T 参数、114 层，采用混合 GDN/MLA 与 MoE（1,224 专家，top-48 + 1 共享）架构及 NVFP4 格式。
- Hacker News front pageAnthropic's Mythos-class model allegedly leaked on Hugging Face
Anthropic 的 Mythos 级模型据称在 Hugging Face 上泄露，相关链接因恶意软件风险已被删除。该帖在 Hacker News 上获得 14 分并引发 4 条评论，目前尚无更多细节披露。
Heat over time
Heat now 9·Comparable peak 10（Nov 2 01:00）·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.