# Everything around the request

> DigitSpace is one API in front of 47 models from 6 providers. These are the parts that work on every request: choosing the model, surviving outages, counting the cost and keeping it within your rules.

Source: https://digitspace.ai/features

- [Smart routing](https://digitspace.ai/features/smart-routing): Send "auto" and DigitSpace picks the cheapest model that can do the job at your quality level, from 47 models and 6 providers.
- [Failover](https://digitspace.ai/features/failover): DigitSpace checks provider health continuously and moves a request to the next provider when one times out, rate-limits you or returns a server error.
- [Cost tracking](https://digitspace.ai/features/cost-tracking): DigitSpace records the model, provider, tokens, cost and latency of every request, with spend by key, member, model, provider and project.
- [Bring your own keys](https://digitspace.ai/features/bring-your-own-keys): Add your OpenAI, Anthropic, Google, Azure, Bedrock or Vertex keys to DigitSpace and pay providers directly. Routing, failover and logs cost nothing.
- [Budgets and access](https://digitspace.ai/features/budgets-and-access): Spending limits per key, member and team, and rules for which models, providers, prices and IP addresses each key may use.
- [Guardrails](https://digitspace.ai/features/guardrails): Guardrails check every request before it reaches a provider and block, redact or flag secrets, personal data, prompt injection, jailbreaks and your own terms.
- [API compatibility](https://digitspace.ai/features/api-compatibility): DigitSpace speaks the OpenAI Chat Completions API and Anthropic's Messages API. Point your SDK at one base URL and use every model.