# stdapi.ai

stdapi.ai is a self-hosted ai app. OpenAI, Anthropic, Cohere, and Ollama-compatible API gateway that routes existing AI applications to 100+ Amazon Bedrock models and AWS AI services (images, speech, embeddings, rerank, batch, Polly, Transcribe, and... It is open source and self-hosted, installable on TrueNAS.

Canonical page: https://ourlocalhost.com/app/4394/stdapi-ai

## Description

OpenAI, Anthropic, Cohere, and Ollama-compatible API gateway that routes existing AI applications to 100+ Amazon Bedrock models and AWS AI services (images, speech, embeddings, rerank, batch, Polly, Transcribe, and Comprehend) in your own AWS account.

## Platform availability

Available on 1 of 10 platforms: TrueNAS.

## Health score

68/100 (Good), based on 4 of 5 factors.
- Maintained: 98 (last commit 52 days ago)
- Popular: 29 (26 GitHub stars)
- Easy to install: 25 (1 install platform, install notes)
- Security check: 100 (no serious findings, CI, tests)

## Links

- GitHub: https://github.com/stdapi-ai/stdapi.ai
- Documentation: https://stdapi.ai/getting_started/
- TrueNAS: https://apps.truenas.com/catalog/stdapi-ai/
- Official website: https://stdapi.ai

## Common questions

### How do I point OpenAI and Anthropic SDKs at stdapi.ai?

Set the OpenAI SDK base URL to your gateway's /v1 path. Point the Anthropic SDK at the /anthropic path on the same gateway.

### How can I reduce Bedrock throttling in stdapi.ai?

Enable additional AWS regions because each region has its own Bedrock quota. stdapi.ai routes and retries eligible requests across enabled regions without changing application code.

### Why does a load balancer health check fail on /health?

Host header validation also applies to /health. If the load balancer probes the target directly, the Host header is the target IP rather than an allowed domain, so probes return 400 and the target group stays unhealthy.

### How is stdapi.ai deployed?

Use the validated Terraform module to deploy it as a 24/7 multi-AZ ECS Fargate service.

### Does stdapi.ai add markup to Bedrock model usage?

Model usage is billed at AWS Bedrock rates with zero markup. The gateway itself costs $0.10 per container-hour.

---
Source: Our Local Host (https://ourlocalhost.com/app/4394/stdapi-ai), the self-hosted and open-source app directory.
