Split test GPT-4 vs Claude
costs automatically
Drop in our proxy API to route prompts between GPT-4 and Claude based on your cost thresholds. Track performance vs spend per model — no code changes needed.
1 API key
replaces both
Real-time
cost tracking
Auto-route
by threshold
Simple pricing
Pro
$29
/month · cancel anytime
- ✓Proxy API key (OpenAI-compatible)
- ✓GPT-4 ↔ Claude auto-routing
- ✓Configurable cost thresholds
- ✓Per-model analytics dashboard
- ✓Performance vs spend tracking
- ✓Up to 5M tokens/month
FAQ
How does the routing work?
You set a cost-per-1k-token threshold. Requests below it go to Claude (cheaper), above it to GPT-4 (more capable). You can also pin specific prompt patterns to a model.
Do I need to change my code?
No. Point your OpenAI SDK base URL to our proxy and swap your API key. We handle the rest — routing, logging, and analytics happen transparently.
What happens to my data?
Prompt content is never stored. We log only token counts, model used, latency, and cost metadata to power your dashboard.