# Run OpenAI & Anthropic Apps on AWS Bedrock Drop-in API gateway for AWS Bedrock and AI services. Build private AI products on AWS — without exposing your data to third-party AI providers, without subscriptions, and without rewriting your applications. Your existing OpenAI and Anthropic applications work immediately — just change the base URL. Access 80+ models with enterprise privacy, compliance controls, and pay-per-use AWS pricing. **14-day free trial included — free for local development.**

[Start 14-Day Free Trial](operations_getting_started.md){ .md-button .md-button--primary } [Try Community Edition](operations_getting_started_local.md){ .md-button }

- :material-swap-horizontal: __Change one line, access 80+ models__
Drop-in replacement for OpenAI and Anthropic SDKs. Works with Open WebUI, n8n, OpenClaw, Claude Code, LangChain, Continue.dev, and 1000+ tools—no code changes beyond the base URL. - :material-shield-lock: __Your data stays in your AWS account__
All inference runs in your account. Data is never shared with model providers or used for training. Configure allowed regions for GDPR, HIPAA, and FedRAMP compliance. - :material-currency-usd-off: __Pay only for what you use__
No subscriptions or monthly minimums. Pay AWS Bedrock rates directly with no markup from stdapi.ai. Pay only for what you actually use. - :material-aws: __Purpose-built for AWS Bedrock__
Deep integration with prompt caching, reasoning modes, guardrails, service tiers, inference profiles, and prompt routers. Not a generic proxy—built to leverage every Bedrock feature. - :material-brain: __Claude, Kimi K2, MiniMax, and 80+ more__
Claude 4.6+ (reasoning), Kimi K2, MiniMax M2.5, Qwen3, Llama 4, GLM 5, Nova 2, Stability AI, and more. Switch models instantly—no vendor lock-in. - :material-rocket-launch: __Deploy in 5 minutes__
3 lines of Terraform for production on AWS. Or run Docker locally for development. Production-ready infrastructure with HTTPS, WAF, auto-scaling, and monitoring included.

```mermaid %%{init: {'flowchart': {'htmlLabels': true}} }%% flowchart LR openai["

OpenAI SDK Apps"] --> stdapi["

stdapi.ai"] anthropic["

Anthropic SDK Apps"] --> stdapi stdapi --> bedrock["

AWS Bedrock
80+ models"] stdapi --> services["

AWS AI Services
Polly · Transcribe · Translate"] ```

**1. Deploy to AWS with Terraform** — use our Terraform module or a managed deployment service. !!! tip "Prefer a hands-off setup?" A [managed deployment service](https://aws.amazon.com/marketplace/pp/prodview-xknxzjgl7zi5s) can deploy stdapi.ai into your AWS account — no Terraform required. **2. Point your application to stdapi.ai** — just change the base URL in your existing OpenAI or Anthropic SDK code. **3. Access any Bedrock model immediately** Use Claude, Kimi K2, MiniMax, or any Bedrock model. Switch between models, regions, and providers without changing application code. **Zero lock-in:** Standard OpenAI and Anthropic APIs mean you can switch back or to another provider anytime.

- :material-earth: __Multiply your quota across regions__
Each AWS region has its own independent quota. Add a second region — double your tokens-per-minute. Add a third — triple it. stdapi.ai routes and fails over automatically; clients never see a throttle error.
[:octicons-arrow-right-24: Resilience & Failover](operations_resilience.md) - :material-star-settings: __Advanced Bedrock capabilities__
Reasoning modes (Claude 4.6+, Nova 2), prompt caching, guardrails, service tiers, application inference profiles, and prompt routers—all through standard OpenAI API parameters. - :material-api: __Complete multi-modal API__
Chat completions, embeddings, image generation/editing/variations, audio speech/transcription/translation. Every route maps OpenAI parameters to Bedrock equivalents. - :material-aws: __Native AWS AI services__
Amazon Polly (text-to-speech), Transcribe (speech-to-text with speaker diarization), Translate—all unified under OpenAI-compatible endpoints. - :material-chart-line: __Full observability__
OpenTelemetry integration for traces and metrics. Detailed request/response logging. Swagger and ReDoc API documentation served by the application. - :material-swap-horizontal: __Automatic deprecated model fallback__
When AWS retires a model, requests are transparently redirected to its replacement. Your applications survive model deprecations without code changes.

- :material-server-network: __DevOps & Platform Teams__
Deploy Open WebUI, LibreChat, or custom chat interfaces for your organization. Unified API gateway for all AI services—no per-application AWS integration needed.
[:octicons-arrow-right-24: Open WebUI guide](use_cases_openwebui.md) · [:octicons-arrow-right-24: All use cases](use_cases.md) - :material-code-braces: __Developers & AI Engineers__
Use Claude, Kimi K2 thinking, and Qwen Coder Next in VS Code (Continue.dev, Cline, Cursor), JetBrains IDEs, or any OpenAI-compatible tool. Test locally with Docker, deploy to production with Terraform.
[:octicons-arrow-right-24: Coding assistants guide](use_cases_coding_assistants.md) - :material-robot: __Workflow Automation Teams__
Connect n8n, Make, Zapier, or custom automation to AWS Bedrock. Access 400+ integrations with enterprise-grade AI—all through one API endpoint.
[:octicons-arrow-right-24: n8n integration guide](use_cases_n8n.md) - :material-domain: __Enterprises with Compliance Needs__
Meet data sovereignty requirements with region controls. GDPR, HIPAA, FedRAMP workloads supported through AWS Bedrock's compliance certifications. CLOUD Act and FISA 702 risk mitigated through customer-managed encryption.
[:octicons-arrow-right-24: Data Sovereignty & Compliance](operations_compliance.md) - :material-cash-multiple: __Cost-conscious Organizations__
Switch from subscription-based AI services to pay-per-use AWS Bedrock pricing. Pay only for actual usage with no monthly commitments while accessing leading models (Claude, Kimi K2, MiniMax, Qwen3). - :material-application-brackets: __Teams Migrating from OpenAI or Anthropic__
LangChain, LlamaIndex, Haystack, Claude SDK, or custom apps work immediately. Gradual migration supported—run both APIs in parallel during transition. - :material-briefcase: __Legal & Professional Services__
Attorneys, consultants, and accountants cannot send client materials to third-party AI services. stdapi.ai processes all inference inside your own AWS account — client data never leaves your infrastructure.

- :material-scale-balance:{ .lg .middle } __Community Edition — Free & Open Source__ --- **Best for:** Open-source projects, local development, testing, and evaluation. - Full API compatibility and all features - Community Docker image - AGPL-3.0 license (source disclosure required for network use) [:octicons-arrow-right-24: Run locally with Docker](operations_getting_started_local.md) - :material-briefcase:{ .lg .middle } __Commercial Edition — AWS Marketplace__ --- **Best for:** Internal tools, SaaS products, proprietary applications, production. - **14-day free trial** — test in your environment risk-free - Hardened container, security updates, commercial support - **$0.10/container-hour** — no markup on model usage; pay AWS Bedrock rates directly - No AGPL restrictions — keep your code and modifications private - Terraform module for production-ready deployment in minutes - Streamlined AWS billing [:octicons-arrow-right-24: Start 14-Day Free Trial](https://aws.amazon.com/marketplace/pp/prodview-su2dajk5zawpo)