Category: Large Language Models

• Skills-Driven Reasoning Paradigm: Tsinghua & Peking University Propose TRS, Saving 59% Tokens Without Accuracy Drop
• Thinking Without Words: Efficient Latent Reasoning with Abstract Chain-of-Thought
• Costs Cut by 90%, Accuracy Hits 100%! MIT's Counterintuitive Architecture Challenges Silicon Valley Dogma
• Can LLMs Enhance Their Own Reasoning? SePT Offers a Simple Online Self-Training Paradigm
• The First Spatio-Temporal Reasoning Framework: Enabling Large Models to Truly Understand Spatio-Temporal Data | ACL'26
• QuantCode-Bench: A Benchmark for Evaluating LLM-Generated Quant Code Quality
• DeepSeek-V4 Preview: Entering the Era of Accessible Million-Token Context
• Official Introduction: Hunyuan Hy3 Preview
• Suspected GPT-6 Revealed! OpenAI Co-founder Breaks Silence on 'Spud': A New AI Model with the 'Smell of a Large Model'! Netizens Comment: It's the First Model That Truly 'Thinks'!
• Xiaomi MiMo-V2.5 Series Large Models Launch Public Beta
• Ordinary Ethernet Cables Can Run Trillion-Parameter Models! Moonshot AI Unveils Breakthrough Architecture: No Need to Buy All H100s! 1T Model Test Shows 64% Latency Drop! The 'Siege' of Large Model Inference is Broken!
• After 10 Weeks of Gains, Global AI Large Model Token Calls Drop for Two Consecutive Weeks: Who Is Paying for the Surge in AI Computing Costs?
• Qwen3.6-Max-Preview is Here!
• NUS, Fudan, and Tsinghua: The First Systematic Survey on Large Model Latent Spaces
• The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
• Anthropic's Research Published in Nature: The Boundaries of LLM Safety Training Are Rewritten
• The World's Most Notorious Forum Uncovered AI's Most Crucial 'Thinking' Ability
• Claude Opus 4.7 Universally Panned! Users Demand a Rollback After Immediate Post-Upgrade Failures: 'Give Me Back 4.6!'
• Long Context Reduced by 60% + 95% Sparsity: A Double Breakthrough Today Sets New Records in Inference Efficiency
• Surging Popularity! MSA Goes Open Source with a Bang!
• The 'Car Wash Dilemma' That Stumped AI Across the Web Has Finally Been Solved
• Large Models Can Now Modify Parameters 'In-Place'! ByteDance Seed & Peking University Paper: Test-Time Inference Requires No Extra Layers or Retraining
• Stanford Confirms: Multi-Agent Reasoning is a Compute Illusion; Single Agents Win with Equal Token Budgets
• Anthropic Unleashes Most Powerful Claude Mythos! Crushes Opus 4.6, Begs Users Not to Use It
• Make Thinking More Accurate and Extended! The New Reinforcement Learning Algorithm FIPO Arrives