How to Check Hermes Agent Token Usage and Cost

Hermes Agent is Nous Research's self-improving terminal coding agent — the one with persistent memory across runs and a sandboxed execution loop. Its headline feature is reach: it can drive 300+ models across a long list of providers, picking whichever fits the step in front of it. Great for results, rough for accounting. Your spend ends up smeared across half a dozen billing dashboards that never add up to one honest number.

How do I see my Hermes Agent token usage?

Hermes keeps a local record of every model call it makes — which model, which provider, how many tokens in and out. You do not have to log into anything to read it. One command scans that record and folds it into a single daily ledger:

$ npx whoburnedmore↳ reading Hermes Agent run history…↳ 1,204 model calls across 6 providers   HERMES AGENT — TOKENS BY DAY  ────────────────────────────────────────  2026-06-17  in 4,118,300  out 312,700  2026-06-16  in 2,640,900  out 201,400  2026-06-15  in 5,902,100  out 470,800   BY PROVIDER  provider A    11.4M tokens  provider B     6.8M tokens  provider C     2.1M tokens   7-day total:  31.7M tokens   est. $128.90

That last line is the one Hermes itself will never print, because Hermes does not know what any single provider charges — it just routes. whoburnedmore fills that gap by reading the run history off disk and attaching a rate to each model it finds, so the total is the sum of real calls rather than a guess.

How do I track cost across Hermes's many models?

This is the part that makes Hermes different from a single-model tool. When one run touches several providers, no provider's own dashboard shows the whole picture — each one only sees its slice. Consolidating means pulling every slice back into one place and labeling it by where it ran 📊:

Notice how the cheapest provider can still carry a big share of calls while a pricier one carries most of the cost — the two are not the same chart. Splitting usage by provider is what lets you ask the useful question: is the expensive route earning its keep, or could the same step go to a cheaper model with no real loss in quality?

Why provider sprawl hides your real number

Suppose a week of Hermes work touched six providers. To total it by hand you would open six billing pages, normalize six different token-pricing conventions, line up six date ranges, and hope none of them rounded differently. Most people never do this, which is exactly how a routing agent quietly becomes the largest line on the bill. Reading one local log sidesteps the whole chore.

How much is Hermes Agent costing me?

Cost is just usage times a rate — but with Hermes the rate is not one number, it is a different price for every model and provider you touched. So the honest total is a sum over calls, each priced by where it actually ran:

We will not quote per-token prices here, because for Hermes there is no single price to quote — rates vary by the model and provider you route to, and they move. What matters is the method: whoburnedmore reads each call, looks up the rate that applied to that specific model, and adds it in. A run that leaned on a premium provider for codegen will cost more than the same token count routed through a budget model, and the total reflects that honestly.

Where does Hermes keep this data, and is it private?

Hermes writes its run history to a local directory on the machine that ran the agent — alongside the persistent memory it uses to improve between sessions. whoburnedmore reads only the numeric side of that record: token counts, model names, provider labels, and timestamps. It does not read your prompts, your code, or your file names, and nothing about them is uploaded. 🛡️

$ npx whoburnedmore --dry-runwould submit (aggregates only):  tokens_in_total   31,742,800  tokens_out_total   2,544,100  models_seen        9  prompts / code / filenames:  never read, never sent

Use --dry-run to print exactly what a submission would contain, or npx whoburnedmore --localto keep the whole breakdown on your own machine and skip the leaderboard entirely. Either way, the agent's memory and your source stay where Hermes left them.