Event ·Multi-source

DeepSeek V4.1 Pro 预训练数据量或突破 50T tokens

一条针对 DeepSeek 下一代模型 V4.1 Pro 的推测认为，其预训练 tokens 规模有望突破 50T。文中提到 Microsoft 在 MAU 上已经用到了 50T，OpenAI 和蚂蚁的用量被认为高于这一数字。作者希望 DeepSeek 的用量远超 50T，并称这将是其首个摆脱算力限制的 Whale 级模型。

Current read

博主在猜 DeepSeek V4.1 Pro 的预训练量会破 50T tokens，还拿 Microsoft 的 MAU 和 OpenAI 的规模做对比，算力党可以看看。

11 sources46° AI heatLast updated 2026/10/22 01:22:16

Evidence

11 sources

Primary sourceTeortaxes

2026/10/22 01:22:16View source

Related source 1Decoder

2026/10/20 18:56:15View source

Related source 2Liputan6 Tekno

2026/10/21 04:11:49View source

Related source 3TestingCatalog

2026/10/20 10:28:39View source

Related source 4Nathan Lambert

2026/10/20 11:43:06View source

Related source 5Artificial Analysis

2026/10/20 16:13:01View source

Related source 6The Information

2026/10/20 19:00:14View source

Related source 7Pandaily

2026/10/21 06:49:49View source

Related source 8Andrew Curran

2026/10/21 14:25:37View source

Related source 9IT之家

2026/10/22 03:00:52View source

Related source 10Tech in Asia

2026/10/20 01:48:07View 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