If your AI agents call models through Amazon Bedrock or Vertex AI, you can now point those calls at condense and let it manage their context. They still run on your own AWS or Google account, with your own credential and the same request format.
Swap your AWS or Google host for one of these, keep the rest of the path, and add your condense key in the x-condense-auth-token header.
https://api.condense.chat/vertex
https://api.condense.chat/bedrock/{region}Vertex went live on September 7. Claude calls to :rawPredict and :streamRawPredict get compressed, and you send your Google access token as usual.
Bedrock followed on September 8. Claude on InvokeModel gets compressed, as do the Anthropic Messages, Chat Completions and Responses endpoints on bedrock-runtime. Streaming works on all of them. You'll need a Bedrock API key sent as a bearer token. SigV4 requests can't go through condense, because the signature covers a body condense rewrites. Bedrock Mantle isn't on this route.
Everything else is forwarded unchanged and isn't compressed, including Bedrock's Converse API and non-Claude models on InvokeModel. Gemini on Vertex was in that group until its own route shipped on September 30.
To manage context, condense compresses the earlier turns of long conversations and adds one short instruction after your system prompt. Short conversations aren't compressed, and responses come back in your provider's format.
The API reference has the full details for Vertex AI and Bedrock.