Here is a summary of the contents within 50% of the total length:

Deep Seek, a company founded by billionaire Leang Wen Fung, is racing to release its next AI model, R2, ahead of schedule. R2 is expected to have improved coding capabilities and be able to reason more effectively in languages beyond English. This move could shake up the industry and undercut Open AI’s prices by 20-40 times.

Meanwhile, Alibaba has announced a new video Foundation model, one 2.1, which outperforms another Open AI model, Sora, on certain benchmarks. The model includes multiple submodels optimized for various tasks, including text-to-video and image-to-video generation.

Open AI, in response, has expanded access to its Deep Research tool, making it available to all paid chat GPT users. They have also released a white paper discussing the potential persuasion risks of their models and are re-evaluating how to test for real-world persuasion.

Finally, Open AI is rolling out a preview of its advanced voice mode, allowing free users to interact with Chat GPT out loud rather than typing.

Translation

Deep Seek 这家公司由亿万富翁梁文峰创立，正在竞相推出其下一个AI模型R2。该模型预计将拥有更好的编码能力，并能在语言之外进行有效的推理。这一动作可能会颠覆整个行业，减少Open AI 的价格30-60倍。 同时，阿里巴巴宣布了一款新型视频Foundation模型，即Alibaba Video 2.1，该模型在某些基准测试中表现更好于另一个Open AI 模型Sora。该模型包含多个子模型，针对不同的任务，包括文本转视频和图片转视频生成。 在此背景下，Open AI 扩大了对其深度研究工具的访问，使之可供所有付费Chat GPT用户使用。此外，他们发布了一份白皮书，讨论了他们模型潜在的说服风险，并重新评估如何测试真实世界的说服风险。 最后，Open AI 推出了其高级语音模式预览版，允许免费用户与 Chat GPT 进行对话，而不是键盘输入。

Reference:

https://www.youtube.com/watch?v=tBG2ZfKYwao

<
Previous Post
Claude 3.7 Sonnet
>
Next Post
Snowflakes CEO Sridhar Ramaswamy interview