Nexus AI Orchestrator — One API. Nine AI models.

What's included

Everything you need. Nothing you don't.

Built on Cloudflare Workers and D1 — globally distributed, serverless, and ready for production from day one.

⚡

Unified endpoint

One POST /chat call to reach any of the nine models. Switch providers by changing a single parameter.

🔄

Automatic fallback

If the primary model fails, the system transparently retries with the next provider in the chain — no downtime, no errors.

🧠

Session memory

Each conversation session persists its last 10 turns in Cloudflare D1. Context follows the user across requests.

🔍

Web search

Enable real-time search on any request with search: true. Powered by Tavily — up to 5 agentic tool calls per response.

💰

Cost tracking

Every call logs token counts and USD cost per model, per session, and per API key. Query totals via GET /usage.

🛡️

EU / US data sovereignty

All integrated providers are hosted in the United States or European Union. No Chinese-jurisdiction infrastructure in the chain.

🔬

Deep Research agent

Autonomous research via POST /research. Gemini plans, searches, and synthesizes multi-source reports in the background — poll for results when ready.

📐

Vector embeddings

3,072-dimension semantic embeddings via POST /embeddings. Single or batch (up to 100 texts). Ready to feed directly into any vector store or RAG pipeline.

Simple by design

One request.
Any model.

Switch between providers without changing your integration. The orchestrator handles routing, fallback, and context automatically.

curl · aiorchestrator.gntkh.com

curl -X POST \
  https://aiorchestrator.gntkh.com/chat \
  -H "Authorization: Bearer orc_..." \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "What happened in markets today?",
    "modelTarget": "gemini",
    "search": true,
    "session_id": "my-session-001"
  }'

# Response includes token counts, USD cost,
# and fallback info if triggered.

Pricing

Pay once. Own your infrastructure.

No subscriptions. You get the source code — deploy it yourself on Cloudflare with your own API keys.

One-time

Starter

Full source code, ready to deploy

^U$D 100

One-time payment — yours to keep

Worker source code (Cloudflare Workers / JS)
Web UI (single HTML file, no build step)
D1 schema + migration scripts
wrangler.toml deploy configuration
9 models: Claude, Gemini, GPT-4.1, Mistral, Grok, Llama, Cohere, Qwen, Perplexity
Session memory, fallback, search, deep research, embeddings, cost tracking
Setup guide + endpoint reference

Get the code

📦 What you receive

ai-orchestrator-worker.js
ai-orchestrator-ui.html
ai-orchestrator-schema.sql
wrangler.toml
README / setup guide

Note: API costs (Anthropic, Google, OpenAI, xAI, Groq, Mistral, Perplexity) and Cloudflare infrastructure are billed directly by each provider to your own accounts. This price covers the code only.

Enterprise

Multi-tenant · custom branding · SLA

Custom

Everything in Starter
Multi-tenant architecture (one worker, multiple clients)
Admin dashboard with billing per consumption
White-label branding
SLA + priority support
Custom model integrations on request

One API. Nine AI models.Zero complexity.