Optimization Articles
Browse 251 articles about Optimization.
What Is DeepSpark? How DeepSeek Made Every LLM 50–400% Faster Without Retraining
DeepSpark is DeepSeek's speculative decoding method that speeds up LLM inference 50–400% with no retraining. Learn how it works and why it matters.
How to Prevent AI Sycophancy: Why Your Agent Agrees With Everything and How to Fix It
AI models agree with users 88% of the time. Learn how to use adversarial councils, devil's advocate prompts, and structured critique to get honest AI feedback.
Static Context vs Dynamic Context in AI Agents: How to Manage What Your Agent Knows
Static context loads every session; dynamic context loads on demand. Learn how to balance both for token efficiency and reliable AI agent performance.
What Is Index Share? How GLM 5.2 Achieves 2.9x Fewer Compute Operations at 1M Token Context
Index Share reuses sparse attention indexers across four layers, cutting compute by 2.9x at 1M token context. Learn how this makes GLM 5.2 affordable to serve.
What Is the AI Token Cost Crisis? Why Uber Burned Its Entire AI Budget in 4 Months
Uber's engineers spent $500–$2,000/month each on AI tokens. Learn why token costs are exploding in agentic workflows and how to manage them.
How to Audit Your AI Agent Harness: 5 Questions to Ask Before Every Model Update
Use this five-question audit to check your agent's sources, reach, job definition, proof requirements, and value before switching to a new model.
Claude Fable 5 Token Costs: How to Manage Usage Without Burning Your Budget
At $10 per million input and $50 per million output tokens, Claude Fable 5 is expensive. Here's how to control costs and get the most from every session.
MCP vs CLI for AI Agents: When to Use Each and Why It Matters for Token Costs
MCP servers load tool definitions into context permanently. CLI tools cost nothing until called. Learn when each integration method is the right choice.
What Is the Slot Machine Method for Claude Code? Why Restarting Beats Correcting
When Claude makes a mistake, arguing makes it worse. The slot machine method says to rewind and re-run instead. Here's why it works and how to do it.
What Is Google Diffusion Gemma? The Text Model That Generates 256 Tokens at Once
Diffusion Gemma uses image generation tech to draft entire paragraphs simultaneously, making it dramatically faster for on-device AI inference.
AI Model Routing in 2026: When to Use Fable 5, Opus, Sonnet, and Haiku
Not every task needs your most expensive model. Learn how to route tasks across Claude Fable 5, Opus, Sonnet, and Haiku to cut costs without losing quality.
How to Use Claude Fable 5 Effort Levels: Low, Medium, High, and Max
Claude Fable 5 has five thinking modes. Learn when to use low vs max effort, why overkill hurts performance, and how to match effort to task complexity.
How to Use the /insights Command in Claude Code to Audit Your AI Workflow
The /insights command generates a 30-day HTML report on your Claude Code usage, surfacing what's working, what's slowing you down, and what to build next.
How to Use Claude Code /rewind to Roll Back Conversations and Code to Any Checkpoint
The /rewind command in Claude Code lets you roll back both code and conversation to an earlier point—better than correcting mistakes mid-session.
How to Use the /status Line in Claude Code to Monitor Context and Model in Real Time
The Claude Code status line shows your model, effort level, and context usage at a glance. Here's how to configure it and why it matters for long sessions.
What Is the Harness vs Model Distinction? Why Your Agent Wrapper Matters More Than Benchmarks
The harness—file access, computer use, concurrency—often drives more performance than the underlying model. Here's how to evaluate both together.
How to Use Prompt Caching to Cut Claude Code Token Costs in Dynamic Workflows
Dynamic workflows burn tokens fast. Learn how to use prompt caching, scope bounding, and Haiku sub-agents to control costs in Claude Code.
How to Manage Token Costs in Claude Code Dynamic Workflows: Haiku Sub-Agents and Scope Bounding
Dynamic workflows can burn millions of tokens fast. Learn how to use Haiku sub-agents, scope bounding, and named deliverables to control costs.
How to Control Token Costs in Claude Code Dynamic Workflows
Dynamic workflows can burn millions of tokens fast. Learn how to scope tasks, use Haiku sub-agents, and set boundaries to keep costs under control.
How to Use Prompt Caching and Token Management in Claude Code Dynamic Workflows
Dynamic workflows can burn through tokens fast. Learn how to use Haiku for sub-agents, bound your scope, and manage costs before they spiral.