82% verified savings. Free until it saves.
Drop-in AI cost optimisation. Swap one URL and CARDIAC-PURR routes every request to the cheapest model that passes your quality threshold — in real time, automatically, with a full audit trail. No model training, fine-tuning, or training data required — it works on day one.
# Before — direct to provider client = openai.OpenAI(api_key="sk-...") # After — through CARDIAC-PURR (drop-in) client = openai.OpenAI( api_key="cardiac-vk-...", ← your key base_url="https://api.cardiac-purr.com/v1", ) # Everything else stays the same.
vs Always-Large
Pure savings-share
OpenAI SDK drop-in
$0 · Zero Risk
How It Works
Three steps to intelligent routing.
Start with zero risk. See your projected savings. Then flip the switch.
base_url to our endpoint. Routing activates. Every request classified in real time — complexity, latency sensitivity, output type — routed to the cheapest model that passes your quality threshold.Pricing
Free until we save you money.
No platform fee. No per-request charge. Zero fee if verified savings are zero.
- Up to 15K requests / month
- Full routing intelligence
- Shadow period included
- Model Delta Invoice
- No credit card required
- Unlimited requests
- 25% → 18% declining share
- $100K monthly hard cap
- You keep ≥75% of savings
- Full MDI audit trail
- Volume-based rate negotiation
- $150K / $200K monthly cap tiers
- SOC 2 Type I roadmap
- Dedicated onboarding
- See enterprise platform →
| Monthly Verified Savings | CARDIAC-PURR Share | You Keep | Monthly Cap |
|---|---|---|---|
| FREE ≤15K req/month | 0% | 100% | — |
| $0 – $75K | 25% | 75% | $100K |
| $75K – $250K | 22% | 78% | $100K |
| $250K – $600K | 20% | 80% | $150K |
| $600K+ | 18% | 82% | $200K |
No per-request fee. No platform fee. Zero fee if verified savings = $0. Tiers applied automatically based on verified monthly savings from your Model Delta Invoice.
Get Access
Start your free account.
Fill the form and confirm your email. API key delivered to your inbox in minutes. No credit card required.
Check Your Inbox
We've sent a confirmation link to . Click it to verify your email — your API key arrives immediately after.
Didn't get it? Resend the confirmation →
base_url="https://api.cardiac-purr.com/v1" api_key="cardiac-vk-..." ← in your confirmation email
Routing accuracy validated on a 100-item gold price sentiment classification benchmark (dataset v60, 15 June 2026) across three provider integrations: Anthropic, OpenAI, and Google. Results: 100% classification accuracy across all three providers (n=100 per provider; 95% confidence interval lower bound: 96.3%). Zero errors recorded. Cost savings versus single-model baseline: 78.8% (Anthropic), 84.9% (OpenAI), 83.5% (Google). Routing distribution: approximately 75% small-tier, 17% medium-tier, 8% large-tier. Azure OpenAI integration excluded from published results pending resolution of provider connectivity issues. Full benchmark methodology and per-item results available in technical documentation upon request.
The CARDIAC-PURR Intelligence Platform — LLM Router has undergone formal risk classification under Regulation (EU) 2024/1689 (EU AI Act). A written Article 6 classification determination has been completed on 25 June 2026 and incorporated into the Article 18 technical documentation.
Article 14 — Human Oversight
Human oversight under Article 14 of Regulation (EU) 2024/1689 requires that natural persons can monitor, interpret, and override automated routing decisions. The CARDIAC-PURR LLM Router provides authorised operators with a real-time operator dashboard and OpenTelemetry-compatible telemetry pipeline through which routing decisions, model selections, provider substitutions, and governance checks can be monitored per request. Operators can review full provenance metadata for every response, identify routing anomalies, and intervene or halt execution through the platform API. A structured audit trail of all routing decisions and operator interventions is maintained automatically.
Automated fallback chains are a separate reliability feature and do not satisfy Article 14 human oversight obligations.
Article 50 transparency obligations apply from 2 August 2026. High-risk AI obligations for standalone Annex III systems apply from 2 December 2027, as amended by the Digital Omnibus on AI (provisional political agreement, 7 May 2026). Enterprises deploying in regulated sectors should contact us regarding compliance documentation.