# Свежие материалы по AI

## Что нового (обновлено: 3 марта 2026)

### OpenAI (официальный RSS)

- [2026-02-28 · Our agreement with the Department of War](https://openai.com/index/our-agreement-with-the-department-of-war)
- [2026-02-27 · Joint Statement from OpenAI and Microsoft](https://openai.com/index/continuing-microsoft-partnership)
- [2026-02-27 · OpenAI and Amazon announce strategic partnership](https://openai.com/index/amazon-partnership)
- [2026-02-27 · Scaling AI for everyone](https://openai.com/index/scaling-ai-for-everyone)
- [2026-02-27 · Introducing the Stateful Runtime Environment for Agents in Amazon Bedrock](https://openai.com/index/introducing-the-stateful-runtime-environment-for-agents-in-amazon-bedrock)

### Google DeepMind (официальный RSS)

- [2026-02-26 · Nano Banana 2: Combining Pro capabilities with lightning-fast speed](https://deepmind.google/blog/nano-banana-2-combining-pro-capabilities-with-lightning-fast-speed/)
- [2026-02-19 · Gemini 3.1 Pro: A smarter model for your most complex tasks](https://deepmind.google/blog/gemini-3-1-pro-a-smarter-model-for-your-most-complex-tasks/)
- [2026-02-18 · A new way to express yourself: Gemini can now create music](https://deepmind.google/blog/a-new-way-to-express-yourself-gemini-can-now-create-music/)
- [2026-02-17 · Accelerating discovery in India through AI-powered science and education](https://deepmind.google/blog/accelerating-discovery-in-india-through-ai-powered-science-and-education/)
- [2026-02-12 · Gemini 3 Deep Think: Advancing science, research and engineering](https://deepmind.google/blog/gemini-3-deep-think-advancing-science-research-and-engineering/)

### Hugging Face (официальный RSS)

- [2026-02-26 · Mixture of Experts (MoEs) in Transformers](https://huggingface.co/blog/moe-transformers)
- [2026-02-24 · Deploying Open Source Vision Language Models (VLM) on Jetson](https://huggingface.co/blog/nvidia/cosmos-on-jetson)
- [2026-02-20 · GGML and llama.cpp join HF to ensure the long-term progress of Local AI](https://huggingface.co/blog/ggml-joins-hf)
- [2026-02-20 · Train AI models with Unsloth and Hugging Face Jobs for FREE](https://huggingface.co/blog/unsloth-jobs)
- [2026-02-12 · OpenEnv in Practice: Evaluating Tool-Using Agents in Real-World Environments](https://huggingface.co/blog/openenv-turing)

### Ещё важные официальные ленты

- Anthropic Newsroom: <https://www.anthropic.com/news>
- Mistral AI News: <https://mistral.ai/news>
- Mistral example update: <https://mistral.ai/news/codestral-25-08>

---

## Что читать в первую очередь (чтобы не утонуть в шуме)

1. **Model releases** (что реально поменялось в качестве/стоимости)
2. **API/platform updates** (новые capability, контекст, tool use, pricing)
3. **Safety/security reports** (ограничения, red teaming, abuse cases)
4. **Benchmarks & evals** (особенно если есть открытая методика)
5. **Infra/ops новости** (runtime, deployment, latency, cost)

---

## Полезные постоянные источники

### Официальные блоги

- OpenAI News: <https://openai.com/news/>
- Google DeepMind News: <https://deepmind.google/blog/>
- Anthropic News: <https://www.anthropic.com/news>
- Mistral News: <https://mistral.ai/news>
- Hugging Face Blog: <https://huggingface.co/blog>

### Исследования и бенчмарки

- arXiv cs.AI: <https://arxiv.org/list/cs.AI/recent>
- arXiv cs.CL: <https://arxiv.org/list/cs.CL/recent>
- Papers with Code: <https://paperswithcode.com/>
- LMSYS Chatbot Arena: <https://lmarena.ai/>

### Практика и экосистема

- vLLM docs: <https://docs.vllm.ai/>
- llama.cpp: <https://github.com/ggml-org/llama.cpp>
- LangChain blog: <https://blog.langchain.dev/>
- Simon Willison AI notes: <https://simonwillison.net/tags/ai/>

---

## Быстрый режим чтения (20 минут в день)

- **5 мин** — заголовки OpenAI/DeepMind/Anthropic/Mistral
- **10 мин** — 1–2 глубоких поста (обычно HF + один вендор)
- **5 мин** — выписать, что влияет на твой стек: модель, цена, latency, инструменты

Если материал не влияет на твои решения по инфраструктуре/продукту — в архив, без чувства вины.