Cheap ModelModel Categories
Understand the workloads Cheap Model is designed to unify.
Text and reasoning
This is the most familiar category for teams migrating from assistant, chat, or coding workflows. It is usually the first place where compatibility and provider choice pay off.
Image generation
Image workloads often need different tradeoffs than text. Teams usually care about prompt fidelity, turnaround time, and commercial output quality more than generic model rankings.
Video generation
Video workloads tend to be slower and more expensive, which makes routing and pricing discipline even more important. They usually deserve their own evaluation path instead of being treated like an extension of image generation.
Audio and speech
Speech recognition, synthesis, music generation, and voice workflows each behave differently. Grouping them under one product surface helps teams compare cost and capability without building separate procurement logic.
Functional APIs
Some workloads need search, crawling, or other data-oriented tools rather than pure generation. Treating these as first-class categories keeps the product useful for agent and operations use cases.
How to choose the right category
- start from the user outcome, not the provider name
- separate evaluation traffic from production-critical traffic
- avoid assuming that one model family should power every modality
- keep routing rules aligned with the job being done