Working notes on running OpenAI-, Anthropic-, Cohere- and Ollama-compatible workloads on Amazon Bedrock: what the catalogue actually contains, what it costs, where it runs, and where the gateway options differ.
On 4 September 2026 AWS deprecated and archived aws-samples/bedrock-access-gateway, the OpenAI-compatible proxy a lot of teams put in front of Amazon Bedrock. The reason given is fair: Bedrock now serves those APIs itself.
Amazon Bedrock quota is per region and per model. If you are seeing ThrottlingException, you are usually not at the limit of what AWS will sell you. You are at the limit of the one region you deployed in. Adding a second region brings that region's own quota with it, and it needs no conversation with your account team.
What if you could deploy a fully private ChatGPT alternative — on your own AWS infrastructure, with your own data sovereignty rules — in 30 minutes?
Use your existing OpenAI-compatible tools (OpenWebUI, n8n, Cline, Cursor) with AWS Bedrock's 100+ models and AWS AI services (Polly, Transcribe). Point the client at your gateway instead of the vendor's. Deploy to your AWS account with Terraform. No third party sits between your users and your models.