Skip to content
Dashboard

Reasoning models vs standard LLMs: Which one should handle each request?

Copy link to headingKey takeaways

Copy link to headingWhat are reasoning models?

Copy link to headingWhat are the differences between reasoning models and standard LLMs?

Copy link to headingWhy reasoning models matter for teams shipping LLM features

Copy link to headingHow reasoning models work

Copy link to heading1. Test-time compute as a second scaling dimension

Copy link to heading2. Chain-of-thought serializes work one forward pass cannot do

Copy link to heading3. Backtracking separates the class from prompted step-by-step output

Copy link to heading4. Effort levels turn the thinking budget into a routing parameter

Copy link to headingWhen to use reasoning models instead of standard LLMs

Copy link to headingWhen the task breaks into checkable steps

Copy link to headingMulti-step math and proofs

Copy link to headingCode debugging against a test suite

Copy link to headingMulti-hop retrieval

Copy link to headingAgentic planning

Copy link to headingWhen a wrong answer costs more than the wait

Copy link to headingWhen escalating backfires

Copy link to headingWhere standard LLMs remain the better default

Copy link to headingThe four routes that belong on a standard model

Copy link to headingWhy format-strict routes are the clearest case

Copy link to headingThe one case where the default is wrong

Copy link to headingFour practices for running a reasoning model route in production

Copy link to heading1. Cap effort before the request reaches the provider

Copy link to heading2. Record hidden reasoning tokens per request

Copy link to heading3. Track time to first token separately from total duration

Copy link to heading4. Move the longest calls off the request thread

Copy link to headingHow Vercel powers reasoning model routing for engineering teams

Copy link to headingAI Gateway routes requests between tiers

Copy link to headingFluid compute runs long reasoning calls

Copy link to headingGateway observability shows what made a request slow

Copy link to headingModel fallbacks keep a route serving

import { streamText } from 'ai';
const result = streamText({
model: 'anthropic/claude-opus-5',
prompt: 'Explain the tradeoffs of server-side rendering',
providerOptions: {
gateway: {
order: ['vertex', 'bedrock', 'anthropic'],
},
},
});
for await (const chunk of result.textStream) {
process.stdout.write(chunk);
}

Copy link to headingRoute reasoning models by task on Vercel

Copy link to headingFrequently asked questions about reasoning models

Copy link to headingDo reasoning models always outperform standard LLMs?

Copy link to headingWhat is the difference between chain-of-thought prompting and a reasoning model?

Copy link to headingWhy is a reasoning model's bill higher than its visible output suggests?

Copy link to headingWhen should a request go to a standard model instead?

Copy link to headingHow long should a reasoning call be allowed to run?

Ready to deploy?