Thinking Models
Thinking Models
Some LLMs improve difficult-task performance by spending more internal computation on problems such as math, coding, and logic.
Summary
Karpathy’s practical advice is to use fast models by default, then switch to a thinking/reasoning model when the problem is hard or the first answer needs deeper work.
When To Use
Use thinking models for:
- Debugging difficult code.
- Math and logic.
- Multi-step reasoning.
- Ambiguous technical diagnosis.
- High-value decisions where latency is acceptable.
Avoid them for:
- Simple recall.
- Basic ideation.
- Low-stakes chat.
- Tasks where extra latency adds no value.
Always Want the Smartest Model
Naval’s selection rule runs the other way from cost-optimization: intelligence is an unalloyed good, so given the leverage poured behind a decision — capital, code, marketing — you want the best judgment every time, and you stop asking the model you believe is weaker once you can’t tell which answer is right. When a model errs you often don’t notice, which is the argument against trading intelligence for price on consequential work.
The caveat comes from Vercel’s gateway data: frontier intelligence at the right cost and latency wins most production traffic, but cheaper and open models earn their place on verifiable, high-volume tasks — support, browser automation — where a wrong answer is caught cheaply. This is the same fast-default / thinking-when-hard logic applied to model choice rather than reasoning depth: spend intelligence where verification is expensive, economize it where checks are cheap. See The AI Industrial Revolution.
Related Concepts
- Human vs AI Capability Lens — the six-model snapshot graded on the five AI facets (which model to spend where).
- LLM Tool Use
- Context Engineering
- Agentic Engineering
- The AI Industrial Revolution
- Global Workspace and J-space — an interpretability view of what extended reasoning does: engaging a deliberate workspace instead of a fast automatic response.
- Automatic and Deliberate Work with AI — when to spend the reasoning budget, as an operator’s routing move.
Sources
- How I use LLMs
- The AI Industrial Revolution
Open Questions
- Which tasks in this wiki deserve a thinking model?
- Should lint passes use a thinking model by default?