What it does
FastRouter gives applications access to models from multiple providers through one OpenAI-compatible API. To start, point an OpenAI SDK client at the FastRouter base URL and use a FastRouter API key. The model field selects the model. The API covers text, image, video, embeddings and speech. FastRouter routes each request by cost, latency or output quality. It provides automatic retries across providers. It reroutes requests to other healthy providers for the chosen models. Teams can define prioritized fallback models and virtual model lists. Dashboards show usage, latency, errors and costs across models and providers.
Capabilities
- Model Council combines multiple models to cross-check outputs.
- Batch processing handles high-volume workloads.
- Alerts cover usage spikes, performance issues, errors and spend anomalies.
- Guardrails validate and monitor inputs and outputs.
- Ongoing evaluations monitor model outputs.
When to use it
- Comparing latency, output quality and cost across models in interactive playgrounds.
- Keeping AI applications running with automatic failover across providers.
- Setting project and API key limits, member roles and access controls.
