Example: CI/CD Budget Gate

Add this to your GitHub Actions to block deploys that use models exceeding your budget:

# .github/workflows/cost-check.yml
name: AI Cost Check
on: [pull_request]

jobs:
  check-cost:
    runs-on: ubuntu-latest
    steps:
      - name: Check model pricing
        run: |
          MODEL="openai-gpt5"
          BUDGET=500  # $500/month max

          COST=$(curl -s "https://www.getapipulse.com/api/calculate?model=$MODEL&requests=10000" | jq '.totalMonthlyCost')

          if (( $(echo "$COST > $BUDGET" | bc -l) )); then
            echo "⚠️ $MODEL costs \$$COST/month — exceeds \$$BUDGET budget!"
            exit 1
          fi
          echo "✅ $MODEL costs \$$COST/month — within budget"

Embeddable Pricing Badges

Add a live pricing badge to your README with one line of markdown:

![GPT-5 Pricing](https://www.getapipulse.com/api/badge?model=openai-gpt5)

This renders an SVG badge showing the model name and current input/output pricing per 1M tokens. It auto-updates when prices change. Works in GitHub READMEs, documentation sites, and Notion pages.

Data Freshness

Pricing data is verified against provider pricing pages regularly. The API response includes a lastUpdated field so you can check freshness programmatically. When providers announce price changes, the data is updated within 24-48 hours.

Models that are deprecated (like Claude 4 Opus, retired June 15, 2026) include a deprecated flag with the deprecation date and replacement model ID, so your tools can handle migrations automatically.

No API Key, No Rate Limits

The API is free to use with no authentication required. It's served from Vercel's edge network with 1-hour caching, so it's fast and reliable. If you're making more than ~1000 requests/minute, consider caching responses locally.

A credit or link to getapipulse.com is appreciated but not required.