以DeepSeek‑R1为例，仅靠强化学习训练，模型在AIME数学推理基准上的pass@1从15.6%提升至 77.9%，充分展示了RL在低数据量条件下即可实现大幅能力跃升，迅速成为后训练赛道的新范式。

新智元报道 ...

The bug allows attacker-controlled model servers to inject code, steal session tokens, and, in some cases, escalate to remote ...

Aider is a “pair-programming” tool that can use various providers as the AI back end, including a locally running instance of Ollama (with its variety of LLM choices). Typically, you would connect to ...

Describing AI development as an "arms race" might seem needlessly bombastic, but there's a reason why this term has entered common usage. It encapsulates the speed and intensity at which companies are ...

OpenAI has gone "code red" multiple times. Sam Altman said he'll sound the alarm again as competition intensifies. OpenAI entered "code red" earlier this month after Google released its latest AI ...

There’s pretty much always a frenzied news cycle of some kind around OpenAI. Fortune’s tech team recently dove into what’s going on behind the scenes, in a feature published this week and helmed by ...

Sam Altman’s decision to declare a “code red” at OpenAI earlier this month may have caught the industry’s attention, but it wasn’t a first for the artificial intelligence company. The San ...

OpenAI needs more compute. It’s not clear what it expects you or anyone else to do about that, but it would very much like you to know that it needs more compute. In a strange video posted by the ...

Anthropic said on Wednesday it would release its Agent Skills technology as an open standard, a strategic bet that sharing its approach to making AI assistants more capable will cement the company's ...