When to Use Reasoning vs Standard Models
Problem types where extended thinking pays off: math, code, multi-step logic.
Not Every Task Needs Reasoning Models
Reasoning models are powerful but expensive and slow. Choosing the right model type for each task is one of the most impactful decisions in LLM system design.
The core question: does this task actually benefit from extended deliberation? Many tasks don't — and using a reasoning model on them wastes money without improving quality.
Where Reasoning Models Shine: Multi-Step Math
Reasoning models dramatically outperform standard models on mathematical problems requiring multiple steps, especially when errors compound across steps.
import anthropic
client = anthropic.Anthropic(api_key='sk-ant-...')
# Multi-step math: use reasoning model
hard_math_prompt = (
'A company has 3 factories. Factory A produces 240 units/day, '
'Factory B produces 180 units/day, and Factory C produces 300 units/day. '
'They operate 5 days/week. A unit sells for $47.50. Operating costs are '
'$18,000/week for A, $14,500/week for B, and $22,000/week for C. '
'What is the total weekly profit across all factories?'
)
response = client.messages.create(
model='claude-opus-4-5',
max_tokens=8000,
thinking={'type': 'enabled', 'budget_tokens': 5000},
messages=[{'role': 'user', 'content': hard_math_prompt}]
)
print(next(b.text for b in response.content if b.type == 'text'))All lessons in this course
- How Reasoning Models Differ
- Effective Prompts for Extended Thinking
- When to Use Reasoning vs Standard Models
- Cost and Latency Tradeoffs