Simple, transparent pricing

Start free. Upgrade when you need enforcement.

Free
¥0forever

Audit + rank + compare — all for free

  • +MCP Server (all tools)
  • +SDK local audit detection
  • +CLI reports & verification
  • +Public ranking access
  • +Community data contribution
  • +API key stays on your machine
Coming Soon
Pro
~¥49/month (estimated)

Proxy mode: audit + budget on every request

  • +Everything in Free
  • +Change base_url, zero code change
  • +Monthly budget enforcement (hard cap)
  • +Per-request cost tracking dashboard
  • +Trust & budget alert notifications
  • +Request history & analytics
  • +Priority support
Coming Soon
Enterprise
Custom

Teams spending $1K+/month on LLMs

  • +Everything in Pro
  • +Private deployment (Docker)
  • +Multi-member budget management
  • +Pay only 20% of actual savings
  • +SLA guarantee
  • +Slack / Feishu integration
  • +Dedicated support

Frequently asked

What does the free SDK actually do?

It hooks into your existing OpenAI SDK calls and passively detects whether the transit provider is serving the model you paid for. Think of it as a trust meter — zero cost, zero risk, your key stays local.

How accurate is the audit detection?

92-97% for large model swaps (e.g., Opus replaced with Haiku), 85-90% for adjacent tiers (Opus vs Sonnet), 95%+ for cross-family detection (Claude vs GPT). Confidence increases with more observations.

Is my API key safe?

With the free SDK, your key never leaves your machine — audit runs 100% locally. With Pro proxy, keys are encrypted at rest (Fernet). Enterprise offers private deployment where keys never leave your infrastructure.

How is ranking data collected?

Three sources: real probe tests (balance-diff verification), automated price crawlers (daily, no keys needed), and anonymized SDK community signals. Higher quality data automatically overrides lower.