Production Routing Simulator

See how ModelSpend
routes every prompt intelligently.

Try our live routing engine. Enter any prompt to see which model (or agent) we'd choose, and the real-world impact on cost, latency and quality. Agentic prompts are detected automatically and routed to reasoning-capable models.

Note: Routing decisions use the production engine. The Answer Preview panel shows an illustrative response. Production routing also applies budget guardrails, DLP scanning, governance rules and provider availability checks.

Enter a prompt or try an example

Please enter a prompt before routing.

Try an example

Live routing result LIVE

Run the simulator to see the live routing result

Works inside your current stack

Your app

Send a prompt from your app

ModelSpend Routing Layer

Evaluates models, cost, latency, quality and policies

Optimal model or agent

Best outcome at the right cost

Response + Usage tracking

Real-time response and analytics

Policy & region aware

Cost & latency optimised

Full observability

Plug in any provider

Live routing engine · Production adds budget, governance, DLP and provider availability checks

Integrate this routing in your app

API Docs →

 # POST /v1/usage — track a routed call curl -X POST https://api.modelspend.best/v1/usage \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d {
    "provider": "anthropic",
    "model": "claude-3-5-sonnet",
    "input_tokens": 152,
    "output_tokens": 320,
    "metadata": { "user_id": "user_123", "workload": "analyst" }
  } 
 from modelspend import ModelSpend

ms = ModelSpend(api_key="$API_KEY")
# Route a prompt and get the optimal model
route = ms.route(
    prompt="Summarise this quarterly report…",
    workload="analyst",
)
print(route.model) # → "claude-3-5-sonnet" print(route.tier) # → "balanced_build" print(route.est_cost) # → 0.072 
 import { ModelSpend } from '@modelspend/sdk';
const ms = new ModelSpend({ apiKey: process.env.MODELSPEND_API_KEY });
// Route a prompt — returns tier, model, and cost estimate const route = await ms.route({
  prompt: 'Summarise this quarterly report…',
  workload: 'analyst',
});

console.log(route.model); // "claude-3-5-sonnet"
console.log(route.tier); // "balanced_build"
console.log(route.estCost); // 0.072 

SDK & tooling

REST API

Easy to integrate

Python SDK

pip install modelspend

Webhooks

Real-time events

Dashboard

Monitor & optimise

View all docs →

See how ModelSpendroutes every prompt intelligently.

Ready to optimise your AI spend?

See how ModelSpend
routes every prompt intelligently.