NEXORA
🐳
🐳
🐳
🐳
🐳
🐳
🐳

About this product

Running your own AI stack should be one docker-compose, not a weekend of yak-shaving. This pack is: Ollama for inference, Open WebUI for chat, LiteLLM as an OpenAI-compatible proxy (so your existing SDK code just works with a base-URL swap), Qdrant for vectors, and Prometheus + Grafana for metrics. Every image is version-pinned. Every config is opinionated + documented. Ships with a model manifest (7 recommended models with RAM requirements), real measured benchmarks on the operator's own hardware, and a hardening guide for the moment you decide to expose it beyond localhost.

What's included

6-service docker-compose with version-pinned images — ollama/ollama:0.6.4, ghcr.io/open-webui/open-webui:0.4.7, ghcr.io/berriai/litellm:v1.52.0, qdrant/qdrant:v1.11.0, prom/prometheus:v2.55.1, grafana/grafana:11.3.0
LiteLLM config with 5 models pre-mapped — llama3.1:8b, llama3.1:70b, mistral:7b, qwen2.5:7b, nomic-embed-text — presents an OpenAI-compatible API on :4000 so any OpenAI SDK works with a base-URL override
Prometheus scrape config wired to Ollama + LiteLLM + Qdrant · Grafana pre-provisioned with Prometheus datasource
Model manifest — 7 model recommendations with Ollama tags + RAM requirements + `docker exec … ollama pull` commands
Real benchmarks (M2 Pro Mac, 32 GB, no GPU): Llama 3.1 8B ~28 tok/s · Mistral 7B ~32 tok/s · Qwen 2.5 7B ~26 tok/s · Nomic Embed ~1200 tok/s — labeled as operator's own measurements, not third-party benchmarks
SETUP.md walks first-run in 7 steps · HARDENING.md covers reverse-proxy + TLS + auth + rate limits + backup before exposing to internet
GPU-ready — commented `deploy.resources.reservations.devices` block; uncomment for NVIDIA acceleration
Credits page lists every upstream OSS license (Ollama MIT, Open WebUI MIT, LiteLLM MIT, Qdrant Apache 2.0, Prometheus Apache 2.0, Grafana AGPL v3) — check model licenses on Hugging Face
Team license: single-org, up to 5 developers · 30-day money-back guarantee

Customer reviews

No verified reviews yet. Be the first to review this product after purchase.

Frequently asked questions

How fast is delivery after purchase?
Instant. Download links and license keys are generated the moment payment clears — typically within 5 seconds. Your email inbox and Library both receive the delivery simultaneously.
Can I re-download later?
Yes. Every purchase lives permanently in your Library. Re-download as many times as your download policy allows — unlimited for most products, lifetime access always.
Do updates cost extra?
No. When the creator ships a new version, your account is upgraded automatically at no cost. You'll see a badge in your Library and receive a notification.
What's the refund policy?
30-day money-back guarantee on all products. If it doesn't work as advertised or isn't right for you, request a refund from your Orders page — approved within 24 hours.
Which payment methods are accepted?
Every major card (Visa, Mastercard, Amex, Discover), Apple Pay, Google Pay, PayPal, and crypto (BTC, ETH, USDC).
You might also like

You might also like