Choosing a model

There is no single perfect model for every task. TokRouter's categories help you pick the right one for your need faster.

Fast

For everyday tasks: rephrasing, short summaries, translation, classification, simple extraction, or conversational replies. These models favor speed and are usually cheaper.

Reasoning

For multi-step problems: complex analysis, math, planning, hypothesis checking, hard debugging, or decisions under constraints. Describe the problem and the expected outcome clearly.

Code

For writing, explaining, reviewing, or fixing code. State the language, version, dependencies, and error message. Share the smallest reproducible snippet possible.

Vision

For images or visual documents. Explain what you expect: text extraction, description, comparison, element identification, or layout analysis.

General-purpose

For long conversations, writing, and mixed tasks that don't fit a specific category.

Compare before standardizing

For an important task, test two models with the same prompt. Compare quality, speed, constraint adherence, and credit usage, then keep the best trade-off.