Resources
Frequently asked questions
Answers about signup, enterprise options, data retention, privacy, and where models run.
See the business email and domain requirements and manual onboarding guidance.
Email partnerships@router.com to discuss Ramp partnership opportunities. Learn more on our Partnerships page.
To explore our custom plans and pricing, talk to our sales team at sales@router.com. Learn more about enterprise options.
Provider-side data retention varies by provider and request path. See the Provider Schedule for zero data retention (ZDR) policies and exceptions. Ramp-side content recording is enabled by default.
To opt out of Ramp-side content recording, disable it at any time in Settings. The preference applies account-wide to your self-service Router API keys and propagates asynchronously, so requests may briefly use the previous setting. Opting out stops future content recording after that propagation window; it does not delete existing archives, and Router may still retain operational metadata such as model, provider, usage, cost, latency, and request status.
When enabled, Router records model inputs, outputs, and tool calls for debugging, semantic analysis, and product improvement. Recorded content is retained for one year by default, and Ramp removes detected personally identifiable information before using that content to improve the product.
Please see the Router Privacy Notice for information on how Ramp manages personal information.
Certain jurisdictions provide rights related to personal data, such as the right to request access, deletion, or correction of the data. If you reside in an applicable jurisdiction and want to submit a data subject request related to your use of Router, you can email privacy@ramp.com with the subject line “Router Data Subject Request” and the details of your request.
For Inspect users not using Ramp SSO, all content recording is turned off and unconfigurable.
Router supports a wide variety of models, including models from frontier labs such as OpenAI, Anthropic, Google Gemini, and xAI, as well as open-weight models. See our Supported models page for a full list.
By default, Router uses shared inference keys. This lets you access every supported model except Google Gemini without any additional setup. To use Google Gemini models, you must provide your own API key. See our BYOK guide for more information. If you don't have a valid Gemini API key, Router will reject requests to Gemini models with a 403 provider_key_required error, and will skip Gemini models when you use multi-model routing.
All models run on U.S.-based infrastructure. Ramp does not host the model weights. Models run in the infrastructure of their underlying provider, such as OpenAI, Anthropic, Google Vertex AI, Fireworks, or xAI. Router is the gateway between your application and those providers: it selects a model, sends the request to that provider, and returns the response in a consistent API format.
The processing region and data-residency options depend on the selected provider and model. You can see which provider handled a request in Logs.
Router supports two API surfaces:
- POST /v1/responses - OpenAI Responses API format
- POST /v1/messages - Anthropic Messages API format
Both surfaces route to the same underlying providers. Responses requests always return an OpenAI Response object, and Messages requests return the Anthropic Messages shape. Router does not serve the OpenAI Chat Completions API; clients on that format should call /v1/responses instead (see Connect for the field mapping).
Run the following command to remove Router configuration and restore your agent's original configuration:
ramp router unconfigureThis removes Router configuration but keeps the Ramp CLI installed.
To remove the Ramp CLI, run:
ramp uninstallSwitchyard routing is an opt-in strategy that chooses between capable and efficient model tiers for each eligible coding-agent turn. The guide explains its signals, eligibility, configuration, and interaction with cost-efficient routing.
Yes. Router supports keys for OpenAI, Anthropic, Google Gemini, Fireworks, xAI, Baseten, and Sail Research. You can add your own provider credentials on the provider keys page, which lists the providers you can add keys for. Amazon Bedrock uses a connection to your AWS account instead of a key; see Bedrock setup.
For Anthropic, use a workspace API key from the Anthropic console.
If a request fails on your key due to an auth error, rate limit, or provider error, Router can optionally fall back to its shared key and bill normally. You can disable this fallback for each provider with an enabled key on the provider keys page.
Models from a provider you have no enabled key for use Router's shared credentials. Google Gemini is the exception: Gemini models require your own key. See What models are supported by Router?. For Amazon Bedrock, see Bedrock billing and fallback.
No. When a request is served with your provider key, your provider bills you directly and Router does not charge for that usage.
If Router falls back to its shared key, that request is billed normally. You can disable shared-key fallback on the provider keys page for each provider with an enabled key. Requests routed to a provider without your enabled key can still use Router's shared credentials.
Why a key gets deactivated. Requests are screened against usage policy categories. A request that scores above the threshold in a category deactivates the key that sent it.
Why several strikes can land at once. Screening runs asynchronously, and coding agents retry automatically. A single prompt that is retried several times can register as several separate violations before you see the first one.
What a ban means. Repeated violations lock the account, and you can't create new keys until an administrator lifts the ban.
Getting unlocked. Contact router@ramp.com. An administrator can lift the ban, and both the ban and the lift are recorded in the audit log.
If you think it was wrong. Say so when you contact us. Screening scores are recorded per request, so a decision can be reviewed rather than argued about.
