Quickstart

First call in 4 minutes.

ModelSpend is an OpenAI-compatible proxy. Change one environment variable — your existing code routes automatically.

1

Get your API key

Create a free account — no credit card required. Then go to Settings → API Keys → Create key.

Your key looks like: msp_live_a1b2c3d4e5f6...

2

Make your first call

# The only change needed — works with any OpenAI-compatible library export OPENAI_API_KEY=msp_live_your_key_here export OPENAI_BASE_URL=https://api.modelspend.best/proxy/v1 # Your existing code is unchanged — run it as normal python your_app.py # or: node your_app.js / npm start / etc.
3

See your savings

After your first call, the dashboard Overview shows live analytics — cost per call, tier distribution, savings vs your original model.

62%
Avg saving
vs routing everything to GPT-4o
< 1 min
Time to insight
after your first call
12+
Provider support
including local Ollama

What ModelSpend just did

🔍 Analysed your prompt complexity and assigned it to a routing tier
⚖️ Checked your budget policies, governance rules, and DLP config
💸 Routed to the cheapest model capable of handling that tier
📊 Logged cost, latency, provider, and business function to your analytics

What to do next

💰
Set a budget
Prevent overspend with company-level caps
🔬
Run an eval
Verify the cheaper model meets your quality bar
📦
Version prompts
Track system prompt changes with the registry
👥
Invite team
Set per-team budgets and governance rules