02WHAT THE MODEL CAN DO, PER TOKEN· SURGING
Models
What the model can do, and what that costs per token. In scope: quantization, mixture-of-experts, distillation, context handling, evals — and the specific-and-local cluster: fine-tuning, personalisation, open-weight releases, and running models on your own compute. General AI research that moves neither cost nor capability per token is out of scope, and the system AROUND the model now lives in `harnesses`.
42SOURCES
27CONCEPTS
10ENTITIES
SOURCE MIX
28 P8 R4 A2 N
ACTIVITY · 20W
010203040506070809101112131415161718192021222324252627
Agent Evaluation Benchmarks
active
Agent Exploitation Attack Surface
active
Agent Harness Evolution
active
Agent Memory Architectures
active
Agent Safety & Alignment
active
Agentic Reasoning
active
Chain-of-Thought Reasoning
active
Circuit Tracing
active
Conditional Memory (Scalable Lookup)
active
Deployed Agent Safety
active
Evolutionary Algorithm Discovery
active
Generative World Models
active
Hierarchical Planning
active
Inference & Token Economics
active
Joint Embedding Predictive Architecture (JEPA)
active
KV-Cache Compression
active
LLM Tool Use
active
Low-Bit Quantization
active
Mechanistic Interpretability
active
Multi-Agent Systems
active
Reinforcement Learning for Agents
active
Self-Supervised Learning (SSL)
active
Speculative Decoding
active
System 2 Reasoning (Objective-Driven AI)
active
Tool-Chain Navigation
active
Vision-Language-Action Models
active
World Models
active