May 15, 2026
Zero users, maximum side-eye
Show HN: Claude Code vs. Codex Global Usage Leaderboard
A shiny AI coding scoreboard drops, and the crowd instantly asks: wait, where’s the data
TLDR: CostHawk launched a public scoreboard for AI coding tool usage, but it currently shows no activity and leans hard on privacy claims. Commenters immediately fixated on a simpler issue: where the numbers come from, with some calling the whole thing polished but confusing.
A new CostHawk leaderboard wants to turn AI coding tools into a public sport, ranking users of Claude Code, Codex, and Cursor by how many tokens they burn through. In plain English: it’s a scoreboard for who’s using these coding assistants the most, with monthly resets, privacy promises, and lots of sleek dashboard energy. There’s just one tiny problem stealing the spotlight: the board shows absolutely nothing yet. Zero users, zero activity, zero drama on the actual chart — so the real action moved straight into the comments.
And wow, the commenters had one main question and they asked it with the persistence of a true internet mob: where is the data coming from? That suspicion popped up immediately and repeatedly, turning the thread into a mini trust crisis. People weren’t arguing over who’s winning the leaderboard; they were squinting at the whole premise. One commenter basically said the page is a perfect example of AI making something look polished while still feeling weirdly empty once you try to understand it. Ouch.
There was also a funny side plot: while some readers were skeptical, another builder slid in with a classic startup-comment move — we’re doing something similar too, but with rewards and private leaderboards. Translation: even in a skeptical thread, someone is already auditioning for the sequel. The result is peak tech-community theater: flashy launch, immediate interrogation, and a crowd more interested in the receipts than the rankings.
Key Points
- •CostHawk launched a public leaderboard that ranks Claude Code, OpenAI Codex, and Cursor users by total token consumption.
- •The leaderboard tracks token counts, models used, sync timestamps, and hashed project identifiers, and says it does not store prompts, responses, or source code.
- •Monthly leaderboard seasons reset at midnight Central Time.
- •The displayed leaderboard snapshot shows no operators, no tracked tokens, and no sync activity yet.
- •CostHawk highlights security and privacy features including AES-256-GCM encryption, local-first telemetry parsing, dry-run previews, and opt-in auto-sync.