DeepSeek V3

Claude AI Co-founder Publishes 4 Big Claims about Near Future: Breakdown

AI Explained

A base AI model compared to its reasoning version, DeepSeek R1.

DeepSeek V3, SGLang, and the state of Open Model Inference in 2025 (Quantization, MoEs, Pricing)

Latent Space

A 671 billion parameter Mixture-of-Experts (MoE) model, considered a leading open-source LLM based on benchmark and chat arena results, notable for its large size and FP8 precision.

"OpenAI is Not God” - The DeepSeek Documentary on Liang Wenfeng, R1 and What's Next

AI Explained

Why Inference Matters 12x More Than Training | YC Paper Club

Y Combinator

A model trained by DeepSeek, requiring 2.8 million H1800 hours for training, highlighting the scale of LLM training.

The Engineering Unlocks Behind DeepSeek | YC Decoded

Y Combinator

DeepSeek's general-purpose base model released in December, comparable to models like GPT-4o and Gemini 1.5. It incorporates various efficiency optimizations and innovations.

Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Enterprise Internal Knowledge

Stanford Online

A model trained on approximately 2.4-2.5 million H800 hours for pre-training.

⚡️Every product of the future will be a living system — Ronak Malde, Trajectory.ai

Latent Space

An advanced open-source AI model from China that significantly impressed the AI community.

Stanford CS329A Self-Improving AI Agents | Part 9 | Future Research Areas

Stanford Online

The base model on which the DeepSeek-Math V2 verification-generation loop is built.