LLM 每日资讯 - 2026-02-14
共采集 418 条，去重排序后精选 Top 10
🎧 语音播报
1. OpenAI has deleted the word ‘safely’ from its mission
📂 hackernews / Hacker News ⭐ 303.0/10 ★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★★
📅 2026-02-13 22:17 UTC
[HN Score: 303, Comments: 185] See also: https://simonwillison.net…
2. Anthropic raises $30B in Series G funding at $380B post-money valuation
📂 hackernews / Hacker News ⭐ 10.0/10 ★★★★★★★★★★
📅 2026-02-12 18:58 UTC
Anthropic获得380亿美元估值和300亿美元G轮融资，创下AI公司融资里程碑。
3. Introducing GPT-5.3-Codex-Spark
📂 blog / OpenAI ⭐ 10.0/10 ★★★★★★★★★★
📅 2026-02-12 10:00 UTC
OpenAI推出首个实时编码模型GPT-5.3-Codex-Spark，速度提升15倍，支持128k上下文。
4. GPT-5.2 derives a new result in theoretical physics
📂 blog / OpenAI ⭐ 9.0/10 ★★★★★★★★★☆
📅 2026-02-13 11:00 UTC
GPT-5.2在理论物理学中推导新公式，后被正式证明和验证，展示了AI在科学研究中的重大突破。
5. Gemini 3 Deep Think: Advancing science, research and engineering
📂 blog / Google AI ⭐ 9.0/10 ★★★★★★★★★☆
📅 2026-02-12 16:13 UTC
Google推出Gemini 3 Deep Think模型，专注于推进科学、研究和工程领域。
6. DeepSight: An All-in-One LM Safety Toolkit
📂 hf_papers / HF Daily Papers ⭐ 8.0/10 ★★★★★★★★☆☆
📅 2026-02-12 10:43 UTC
DeepSight提供大模型安全评估、诊断和对齐的一体化工具，解决了当前LLM安全工作流中工具分散的问题。
7. DeepGen 1.0: A Lightweight Unified Multimodal Model for Advancing Image Generation and Editing
📂 hf_papers / HF Daily Papers ⭐ 8.0/10 ★★★★★★★★☆☆
📅 2026-02-12 12:44 UTC
DeepGen 1.0提出轻量级5B统一模型，实现图像生成和编辑的综合能力，解决了大参数规模模型的高成本问题。
8. Pretraining A Large Language Model using Distributed GPUs: A Memory-Efficient Decentralized Paradigm
📂 hf_papers / HF Daily Papers ⭐ 8.0/10 ★★★★★★★★☆☆
📅 2026-02-11 23:02 UTC
提出分布式GPU预训练大语言模型的高效去中心化范式，解决了传统集中式训练的高内存需求问题。
9. [HF Model] mistralai/Devstral-2-123B-Instruct-2512
📂 hf_models / mistralai ⭐ 8.0/10 ★★★★★★★★☆☆
📅 2026-02-12 13:31 UTC
MistralAI发布Devstral-2-123B-Instruct-2512指令模型，属于行业重要模型发布。
10. [HF Model] nvidia/Cosmos-Transfer2.5-2B
📂 hf_models / nvidia ⭐ 8.0/10 ★★★★★★★★☆☆
📅 2026-02-12 22:42 UTC
NVIDIA发布Cosmos-Transfer2.5-2B模型，支持文本到视频和图像到视频生成，属于行业重要模型发布。