Products¶

  Drop-in replacement

Chat, embeddings, reranking, images, video, audio, files, and moderation — 50+ endpoints across three protocols. Standard OpenAI, Anthropic, and Cohere SDKs connect on the base URL alone.

  100+ models

Claude, OpenAI GPT, xAI Grok, Kimi, DeepSeek, Qwen, GLM, Nova, Llama, and more — from Anthropic, OpenAI, xAI, Moonshot, DeepSeek, Alibaba, Zhipu, Amazon, and Meta. Bedrock, Bedrock Mantle, Polly, Transcribe, and Comprehend surface as one catalogue, discovered automatically. Switch models on the fly — no vendor lock-in.

  No third party between your users and your models

The gateway runs in your own AWS account, and Amazon Bedrock does not share prompts with model providers or use them for training. Configure region allow-lists to match your data-residency requirements — AWS compliance certifications apply to the AWS services and regions you choose, and are not inherited by stdapi.ai.

  One quota per region, and regional retry

Every AWS region has its own Bedrock quota, and each region you enable adds its own. Eligible throttling and availability failures retry in another enabled region — streaming retries only before the stream opens, and asynchronous jobs stay in the region that accepted them.

  Pay only for what you use

0% markup on model usage — Amazon Bedrock is billed to you directly by AWS, at AWS rates. The gateway licence is $0.10 per container-hour on the AWS Marketplace (14-day free trial, $0.09 through a private offer), or $0 with the AGPL-3.0 Community edition. The Terraform default runs one task per availability zone — about $216/month in a three-AZ region.

  Advanced Bedrock features

Reasoning modes, prompt caching, guardrails, inference profiles, and prompt routers — all through standard API parameters.