NewOpenTier v1.1.1 is now available

Intelligent Knowledge

Built for Action

OpenTier turns organizational knowledge into real-time, actionable intelligence. It's secure by design, fast by default, and built to scale with your business.

OPENTIER SYSTEM
How does OpenTier scale?

SYSTEM OUTPUT · SIMULATION MODE

Scalable Intelligence

OpenTier combines the performance of Rust with the intelligence of Python,
creating a production-ready platform that scales with your needs.

Client
Rust Gateway
Python Engine
Redis 8 Streams
Worker Engine
Qdrant Vector DB
PostgreSQL 18

Control Gateway (Rust)

Handles public API traffic, authentication, tiered rate limits, and SSE streaming with zero-cost abstractions and sub-millisecond overhead.

gRPC & Redis Backbone

Strongly-typed Protobuf RPCs connect services while Redis 8 Streams coordinates decoupled events with zero loss guarantees.

Cognition Engine (Python)

Orchestrates multi-provider LLM inference, conversation history, query expansion, and hybrid retrieval isolated from public traffic.

Hybrid Vectors (Qdrant)

System of record for vector embeddings, executing sub-millisecond dense cosine search fused with sparse BM25 lexical tokens.

Autonomous Workers

Dedicated stream consumers handle heavy background tasks: web scraping, token chunking, embedding generation, and billing metering.

ACID System of Record

PostgreSQL 18 stores users, organizations, permissions, document metadata, and immutable credit transaction ledgers.

Rust-Powered Gateway

Blazing-fast API gateway built with Axum for maximum throughput, memory safety, and native SSE streaming.

Qdrant Hybrid Retrieval

Dense neural embeddings (up to 3072d) fused with sparse BM25 lexical tokens via Reciprocal Rank Fusion.

Redis Event Backbone

Redis 8 AOF event streaming with consumer groups, distributed locks, SHA-256 deduplication, and DLQs.

Autonomous Workers

Decoupled background workers executing long-running web scraping, document chunking, and billing metering.

Enterprise Security

OAuth 2.0, AES-256 encrypted provider keys at rest, session management, and granular RBAC built-in.

Multi-Model Intelligence

Seamlessly route across Google Gemini, OpenAI GPT, and local Ollama models with custom endpoint overrides.

Real-Time Streaming

Server-sent events and gRPC server streaming for instant, responsive token-by-token chat experiences.

Production Ready

Docker Compose orchestration, self-healing workers, built-in observability, and zero-downtime migrations.

Powered By Modern Technologies

Tokio
SQLAlchemy
Tokio
SQLAlchemy
Tokio
SQLAlchemy
Tokio
SQLAlchemy

Up and running in 3 steps

Download the production compose file, deploy the published GHCR containers, and stream intelligence.

01

Fetch Compose & Config

Download the production compose file and environment template — no source code cloning required.

curl -O https://raw.githubusercontent.com/Celestial-0/OpenTier/main/server/docker-compose.yml
curl -O https://raw.githubusercontent.com/Celestial-0/OpenTier/main/server/.env.example
cp .env.example .env
02

Deploy with Docker

Pulls published GHCR container images and boots PostgreSQL 18, Redis 8, Qdrant, Workers, and Gateway.

docker compose up -d

# Pulls from GitHub Container Registry (GHCR):
# - ghcr.io/celestial-0/opentier-api:latest
# - ghcr.io/celestial-0/opentier-intelligence:latest
# + PostgreSQL 18, Redis 8 Streams, and Qdrant
03

Query the API

Hit the local Rust gateway directly to execute hybrid vector searches and stream real-time tokens.

curl http://localhost:4000/v1/chat \
  -H "Authorization: Bearer <token>" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarize the docs"}'

Ready to deploy?

Read the full self-hosting guide or browse the source on GitHub.

Common Questions

Everything you need to know about integrating and deploying OpenTier.

Get in touch

We'd love to hear from you

Have a question, feedback, or partnership inquiry?
Drop us a message and our team will get back to you.

Get in touch

Send us a message

Have a question, enterprise inquiry, or just want to say hi? We typically respond within 24 hours.

Enterprise & Partnerships

Custom deployments, SLAs, and dedicated support.

Open Source Contributions

PRs, issues, and RFC discussions welcome on GitHub.

Technical Questions

Integration help, architecture reviews, and debugging.