Features Pricing Docs About
DOWNLOAD

Install in 15 minutes.
Your cluster, your rules.

Dream-Weaver ships as a Helm chart, Docker Compose stack, or standalone tarball. Open Core is free forever. Pro features unlock with a license key.

✓ v0.9.1-beta ✓ Linux / macOS / Windows WSL2 ✓ Helm 3.x & Docker 24+

  1. Prerequisites

    • Kubernetes cluster (k3s, k3d, EKS, GKE — anything)
    • Helm 3.x installed: brew install helm or see helm.sh
    • Ollama running on a node with a GPU (optional but recommended): ollama serve

  2. Install the chart

    $ helm install dream-weaver \ oci://ghcr.io/poser8/dream-weaver \ --namespace dreamweaver \ --create-namespace \ --set global.apiKey=dw-live-YOUR_KEY_HERE

    Installs proxy, memory (pgvector), and Ollama side-car. Ready in ~2 minutes.

  3. Verify the install

    $ kubectl -n dreamweaver rollout status deploy/proxy && \ curl http://$(kubectl -n dreamweaver get svc proxy -o jsonpath='{.spec.clusterIP}')/health

    ✓ Expect: {"status":"healthy","version":"0.9.1"}

  4. Point your OpenAI SDK at it

    app.py
    from openai import OpenAI client = OpenAI( api_key="dw-live-YOUR_KEY_HERE", base_url="http://proxy.dreamweaver.svc.cluster.local/v1" )

  5. Enable Pro features (optional)

    If you have a Pro license key, pass it at install time to unlock auto-routing, semantic memory, and the LSR reasoning pod:

    $ helm upgrade dream-weaver oci://ghcr.io/poser8/dream-weaver \ --namespace dreamweaver \ --set global.apiKey=dw-live-... \ --set license.key=DW-PRO-XXXX-XXXX-XXXX

  1. Prerequisites

    • Docker 24+ and Docker Compose v2
    • At least 8GB RAM recommended (16GB for models)
    • Optional: NVIDIA GPU with nvidia-container-toolkit for local inference

  2. Download and run

    $ curl -O https://releases.dream-weaver.ai/latest/docker-compose.yml && \ DW_API_KEY=dw-live-YOUR_KEY_HERE docker compose up -d

    Starts: proxy (port 8080), Postgres+pgvector, Redis, and optional Ollama side-car.

  3. Test the endpoint

    $ curl http://localhost:8080/health

  4. Add a Pro license key (optional)

    Set DW_LICENSE_KEY=DW-PRO-XXXX-XXXX-XXXX in your environment or .env file and restart.

  1. Prerequisites

    • Python 3.10+ and pip
    • Postgres 15+ with pgvector extension
    • Redis 7+
    • Optional: Ollama for local model inference

  2. Download the release

    📦
    dream-weaver-0.9.1-beta.tar.gz FREE
    Full source + pre-built proxy binary. ~18MB. SHA-256 verified.
    Download

  3. Install and configure

    $ tar -xzf dream-weaver-0.9.1-beta.tar.gz && cd dream-weaver && \ cp .env.example .env && nano .env # set POSTGRES_DSN, REDIS_URL, API_KEY

  4. Run

    $ pip install -r requirements.txt && python -m dream_weaver.proxy
PRO LICENSE

Unlock Pro features

After subscribing, your license key arrives by email. Enter it below to reveal your key and full setup instructions. Or paste it directly into your Helm values / .env file.

I have a license key

Enter your license key to reveal your Pro download assets and feature activation commands. Don't have one yet? Subscribe to Pro →

✓ License key valid — Pro features unlocked

Setup guide →

Quick activate:

$ helm upgrade dream-weaver ... --set license.key=DW-PRO-...
WHAT'S INCLUDED

Open Core vs Pro

🔓
Open Core FREE
Full OpenAI-compatible proxy, Ollama/local GPU inference, explicit model selection, streaming, function calling, REST API, community support. Self-host forever at no cost.
Install above ↑
🧠
Pro $49/mo
Auto-routing MoE classifier, persistent semantic memory (pgvector + mxbai-embed-large 1024-dim), LSR neuro-symbolic reasoning pod, multi-tenant API keys, audit log + cost attribution, policy-based routing rules, MCP tool server, email support.
Subscribe →
🏢
Enterprise Custom
Air-gapped VPC deploy, SSO/SAML/LDAP, SLA, dedicated cluster, custom model integrations, white-glove onboarding, dedicated Slack channel.
Contact us →
REQUIREMENTS

What you need

MINIMUM (proxy only)
  • 🖥 2 vCPU, 4GB RAM
  • 💾 8GB disk
  • 🐋 Docker 24+ or Kubernetes 1.26+
  • 🌐 Any NVIDIA/AMD/CPU host
RECOMMENDED (with local inference)
  • 🖥 8+ vCPU, 32GB RAM
  • 🎮 NVIDIA RTX 3080+ or AMD MI200+ (12GB+ VRAM)
  • 💾 100GB+ NVMe (models are large)
  • 🌐 Gigabit LAN or better

No local GPU? Cloud routing still works — you just won't get the local inference cost savings. Local GPU is recommended for the 60–80% token cost reduction.

GET STARTED

Stuck? We've got you.

Read the quick-start docs, or email us. Pro subscribers get email support with same-day response.

Quick-start guide → Email support