groundy
articlessearch

models & research

  1. How LLMs Track Who Did What: The Entity Rebinding Circuit
  2. Claude Fable 5 vs Opus 4.8: When 2x Pricing Is Worth It
  3. Claude Mythos 5 Access Rules: Who Gets Project Glasswing and Why
  4. Fable 5 Distillation Protection: How Anthropic Blocks Model Copying
  5. Can LLMs Write Better Research Paper Titles Than Authors?
  6. Does Information-Theoretic Example Selection Beat kNN for In-Context Learning?
  7. Do Concept Bottleneck Model Benchmarks Measure Interpretability or Dataset Bias?
  8. Persona Prompts Change Who an LLM Recommends as an Expert
  9. Opus 4.8 Batch API: 1M Context, 300k Output, and Team Cost Controls
  10. μP Hyperparameter Transfer Has an Embedding Layer Hole, New arXiv Paper Says
  11. Qwen3.6-27B's Dense Architecture Challenges the MoE-Only Playbook for Flagship-Class Coding Models
  12. Chinese AI Models Compared: DeepSeek, Qwen, Kimi, Doubao, and Ernie
  13. Running DeepSeek R1 Locally: Hardware Requirements, Quantization, and Real Throughput
  14. Fish-Speech: The Open-Source TTS Model That's Threatening ElevenLabs
  15. Synthetic Data Is Eating AI Training
  16. Google's TimesFM: A Foundation Model for Time Series
  17. Gemini 2.0 Pro's 2 Million Token Context: What Can You Actually Do With It?
  18. DeepSeek V3/R1: How Chinese Engineers Matched GPT-4 for $6 Million
  19. Claude's Web Search Changes Everything for AI Research
  20. The Million-Token Context Window: What Can You Actually Do?
  21. Kimi Claw: Moonshot AI's Answer to Claude and ChatGPT
  22. WiFi DensePose: Full-Body Tracking Through Walls Using Your Router
  23. AI Code Generation Benchmarks 2026: Which Model Actually Writes Better Code?
  24. The Best AI Models for OpenClaw in 2026