Skynet Adds Smart Routing Across Every Frontier Model
SAN FRANCISCO: Skynet today shipped smart routing, which selects the model for each prompt automatically, balancing quality, speed, and cost on every request.
Choosing a model has been an unhelpful question to put in front of users. The best model for summarising a long document is rarely the best one for writing code, and neither is the right choice for a one-line question that any fast model answers correctly for a fraction of the cost.
Smart routing removes the question. Each prompt is assessed and sent to the model best suited to it across GPT, Claude, Gemini, and others. Users see one chat; underneath, the workload is spread across the providers that serve it best.
Why this lowers the bill
Routing is also the cost mechanism. Simple prompts do not need frontier reasoning models, and the majority of everyday prompts are simple. Sending them to a fast model and reserving the expensive ones for work that genuinely benefits keeps quality where it matters and spend where it does not.
Smart routing is on by default for all paid plans, and available on the free plan across a curated set of fast models.