Model Proxy is a configuration and validation console for multi-provider AI setups. Generate client configs for Claude Code, Codex, OpenAI SDK, and Anthropic SDK; configure provider profiles, model aliases, and fallback candidates; and simulate routing and budget outcomes. By default, telemetry retains only metrics, not prompt or response bodies.
Hey Product Hunt! I built Model Proxy after repeatedly seeing multi-provider AI setups split across client configs, aliases, fallback rules, and budget spreadsheets.
This first release focuses on making those decisions visible before real traffic: generate configs for Claude Code, Codex, and popular SDKs; configure providers, aliases, and fallbacks; then simulate routing and budget outcomes.
Privacy is deliberate: the default analytics model retains metrics only, not prompt or response bodies.
If you work across several providers, I’d love to know: what’s the hardest part of keeping client configuration, routing, and cost control consistent?