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