LLMs Research covers large language model research papers for people who build with LLMs professionally. Roughly 100 relevant papers hit arXiv each week. Each issue filters that volume down to what changes practice, explains the mechanism rather than the abstract, and states plainly what is hype and what is usable now. Coverage sits in four areas: reasoning and agents (planning, tool use, multi-agent coordination), efficiency (quantization, inference optimization, distillation, KV cache), architecture (attention variants, memory systems, MoE, scaling behavior), and applied systems (RAG, code generation, evaluation). The lab also maintains multiple public repositories of paper implementations, including PaperBanana (2.3k+ stars), and publishes peer-reviewed research at reputed conferences.