Models · 2026-08-28 · HippoAPI Documentation Team
Choose and switch models
Use task quality, modality, context, price, and availability to shortlist the right model identifier.
Start with task requirements
Define required input and output modalities, context size, structured output or tool needs, acceptable latency, and cost range before comparing model names.
Evaluate representative work
Build a small test set from real, authorized examples. Compare answer quality, failure rate, latency, and total usage cost across a short list rather than relying on one benchmark.
Switch with configuration
Keep the model identifier and model-specific parameters in application configuration. When switching, re-run the evaluation set and confirm context, tool, safety, and output-format behavior.