In depth
Frontier models are what you reach for when quality matters more than cost: writing the actual code, producing the final user-facing answer, planning a complex task. They are typically 10× to 30× more expensive per token than the same provider's fast tier. Routing them to the right turns is the cost trick.
Related concepts
LLMA neural network trained on text that takes a prompt and returns text, optionally including structured tool calls.Open-weight modelA model whose weights are publicly downloadable, runnable on your own hardware via Ollama, vLLM, or similar.Model routingSending different turn types to different models (cheap for exploration, premium for writing).
Read the deep dive
How we cut our coding agent's bill by 60% with model routing
Read article
Newsletter
Get the next post in your inbox.
Engineering notes from the Digitorn team. No marketing, no launch announcements, no "10 prompts that will change your life". Just the things we write that we'd want to read.