A curated archive of frontier LLM, multimodal, reasoning, and vertical model technical reports — papers, system cards, and official documentations organized by year and lab.
Latest Technical Reports
See all reports
DeepSeek-R1: Incentivizing Reasoning in LLMs via Reinforcement Learning
Demonstrates remarkable reasoning capabilities via large-scale reinforcement learning without supervised fine-tuning as a preliminary step.
Claude Sonnet 5: System Card & Frontier Evaluation
Clear gains over Claude Sonnet 4.6 in coding, agentic search, multimodal reasoning, and professional-task performance across benchmarks.
GPT-5.4 Thinking: Frontier Reasoning & Deep Research
Unifies recent gains in coding, agentic workflows, and deep web research, while adding high-capability cybersecurity mitigations.
Milestone Evolution Timeline
View categorized catalogCategorized Catalog
Technical reports categorized by core capability, architectural paradigm, and vertical focus.
Technical Reports Archive
Reports Leaderboard Matrix
112 Reports| # | Model Name | Organization | Release Date | Core Technical Highlights | Modalities & Tags | Links |
|---|
Labs & Organizations Tracked
Curating primary technical documentation from the world's leading AI research institutions.
OpenAI, Anthropic, Google DeepMind, DeepSeek, Alibaba (Qwen), Zhipu AI, Moonshot AI (Kimi), Tencent Hunyuan, ByteDance, MiniMax, Meituan (LongCat), Meta AI, NVIDIA, Microsoft, StepFun, InternLM, OpenBMB, InclusionAI (Ant Group), Allen AI, Hugging Face, Snowflake, xAI.