Loading rankings…
Loading rankings…
Yield (Υ) vs total token consumption — efficiency vs volume
| Feature | SigRank | CostHawk |
|---|---|---|
| What it measures | Token-cascade efficiency (Yield, Leverage, SNR, Velocity) | Total token consumption (anonymized) |
| Headline metric | Yield (Υ) = (cache_read × output) / input² | Total tokens consumed |
| What it tells you | How efficiently you use AI — are tokens compounding? | How much you spent — volume only |
| Leaderboard | Yes — ranked by Yield, class tiers, weekly drops | Yes — anonymized, ranked by consumption |
| Class tiers | Yes — 8-tier experience ladder | No |
| Build archetypes | Yes — 10 build archetypes | No |
| API | Public REST API + MCP tools | Limited |
| Privacy | Token counts only — never prompt content | Token counts only |
CostHawk is privacy-first with an anonymized leaderboard. If you want to see total token consumption without attaching your name to it, the anonymized approach is a legitimate privacy choice.
But total consumption doesn't tell you if you're good at using AI. An operator who burns 10M tokens to produce 1K output has high consumption — but low efficiency. An operator who uses 100K tokens to produce the same 1K output has lower consumption but 10x higher Yield. Consumption rewards volume. Efficiency rewards skill.
SigRank's Yield (Υ) metric measures the architecture of your token cascade — is signal compounding through cache reuse and tight output, or are tokens burned in long input chains? CostHawk tells you how much you consumed. SigRank tells you whether those tokens were well spent.
The key difference: CostHawk ranks by consumption. SigRank ranks by efficiency. Both respect privacy. Only one measures skill.
CostHawk ranks by total token consumption — how much you burned, anonymized. SigRank ranks by Yield — token-cascade efficiency. Consumption rewards volume; Yield rewards skill. An AI user who burns 10M tokens for 1K output ranks high on CostHawk but low on SigRank.
No — both respect privacy (token counts only) but answer different questions. CostHawk tells a developer how much they consumed. SigRank tells them whether those tokens were well spent. They're complementary for a person using AI who wants both volume and efficiency views.
Yield (Υ) = (cache_read × output) / input² — token-cascade efficiency from real sessions. Works across Claude, GPT, Gemini, Cursor, Copilot, and any AI coding tool.
Visit signalaf.com/score to enroll and submit your token telemetry. SigRank will compute your Yield, your rank, and your operator class. Token counts only — never prompt content, never code.
Stop consuming. Start measuring.
Check my Yield