LLM Model Selection Skill

LLM model selection evaluates task complexity to map requirements to model tiers like Frontier, Capable, or Fast. It balances capability, cost, and latency to identify the most suitable tier for development workflows and long-context conversations.

You should escalate to a more capable LLM model tier when handling complex multi-file updates or long-context tasks. Escalating ensures higher capability models are reserved for architecture decisions where lower tiers would produce suboptimal results.

The best way to balance cost and latency is applying a tiered model-rating framework that categorizes options by power and speed. This approach informs architecture decisions by matching task types to optimal cost profiles and latency requirements.

To switch between LLM model tiers in a multi-turn chat, you apply practical switching steps based on the assessed task complexity. This involves escalating to more capable models for long-context tasks while dropping to faster tiers for simpler queries.

Yes, the model selection approach works for multi-file code updates by providing guidance for architecture decisions. It helps decide whether to escalate to more capable models like Claude Opus 4.5 versus GPT-4o based on the task's complexity and context length.