burnlog vs viberank

This is the real overlap: viberank is a community leaderboard for AI coding usage, ranked by cost and tokens, with a tiered ladder, GitHub-verified submissions, per-tool filtering and shareable rank cards. If you want a board to be on, it is a good one and it got there first.

The differences worth knowing are three. viberank's submissions are built from ccusage output or an uploaded cc.json, so its coverage is whatever ccusage can read; burnlog also counts at the wire, which reaches agents that write no log at all. viberank ranks on cost as well as tokens, which means uploading what you spent; burnlog uploads token counts only and estimates cost server-side, so your spend stays your business. And burnlog's ladder runs eleven ranks to ten trillion tokens, because real usage blew through a ten-million ceiling long ago.

what viberank is

A community leaderboard for AI coding usage. Data is submitted with its own CLI, which reads local ccusage data, or by uploading a cc.json; GitHub sign-in verifies a submission.

MIT · github.com/sculptdotfun/viberank

Side by side

	burnlog	viberank
Ranked by	Tokens burned	Cost and tokens
What you upload	Token counts, model, agent, opaque id, timestamp	Total API costs, token counts, and which tools contributed
Where numbers come from	Own log adapters, plus the wrap proxy reading provider usage fields	ccusage output, or a cc.json you upload
Agents with no readable log	Counted through wrap, across 14 providers	Limited to what ccusage can read
Rank ladder	11 ranks, Spark to Boltzmann, topping out at 10T tokens	Tiered ladder, Spark through Supernova, based on spend
Verification	GitHub sign-in; events deduped by an opaque per-request id	GitHub sign-in, with verified badges and server-side validation
Self-hosting	Whole app runs against your own Postgres	Source is MIT-licensed
Also has	Achievements, README badge, embeddable widget, MCP server	Per-tool leaderboards, rank cards, a hiring marketplace, a blog

viberank facts read from its own README and site on 2026-08-13. If something here is out of date, tell us and it gets fixed.

Use viberank when

→Your usage is Claude Code through ccusage, and you'd rather be ranked on what you spent than on what you burned.

→You want the hiring marketplace and the per-tool boards it runs alongside the leaderboard.

→You're already there with history you don't want to leave behind.

Use burnlog when

→You'd rather not upload what you spend. burnlog never receives a cost figure; it estimates one from tokens, per bucket, and cache reads are priced as cache reads.

→Your burn isn't only Claude Code: wrap counts Gemini CLI, aider, opencode, your own scripts, and anything else that reads a standard base-URL variable.

→You're past the top of a short ladder and want a rank that still means something at eleven figures.

Try it in a minute

It scans, shows you your numbers, and only then asks for an account. Nothing to uninstall if you don't like it.

$ npx @sxnalabs/burnlog

See the leaderboard →

other comparisons

vs ccusage vs CCgather