Hyperstrate v0.1.0

Hyperstrate v0.1.0

Initial public release of Hyperstrate: a self-hosted AI gateway runtime and control plane for routing, governing, and observing production LLM traffic.

What’s included

• Multi-provider AI gateway runtime
• OpenAI-compatible, Anthropic-compatible, native, streaming, async, and proxy inference flows
• Router pipelines with retry, fallback, caching, rate limits, budgets, quality gates, semantic memory, and MCP tools
• Interceptors for semantic routing, A/B testing, content filtering, PII handling, prompt guard, prompt shield, and team budgets
• Virtual keys, API keys, teams, users, organizations, and OIDC/JWKS group mappings
• Inference logs, pipeline traces, analytics, Prometheus metrics, audit logs, CSV export, webhooks, and request replay
• Prompt templates with versioning
• Evaluation sets and router regression runs
• Vue control plane for managing providers, routers, prompts, evals, MCP tools, teams, keys, and analytics
• Local SQLite development mode and AWS Lambda/PostgreSQL/SQS deployment path

Try it locally

Start the server, then run the client and configure the API endpoint.

Good for

• Teams building production LLM apps
• Developers who want one gateway in front of multiple model providers
• Teams that need budgets, rate limits, evals, traces, virtual keys, and model/provider governance
• Self-hosted AI infrastructure setups