Track AI Token Spend in Grafana: Claude, Codex, and Ollama

A developer built a monitoring setup to track AI token spend and usage across Claude Code, Codex, and Ollama in a homelab, integrating metrics into Prometheus and Grafana. The system handles three different billing models—subscription quotas for the CLIs and per-token API costs for an automated agent—using OpenTelemetry, a custom Python exporter, and Pushgateway. The developer also documented PromQL pitfalls and privacy scrubbing for publishable dashboards.

I let three different AIs work in my homelab every day — a coding assistant, a second CLI for pair-work, and a small agent that triages alerts overnight. One evening I realized I couldn't answer a basic question: what is all of this actually costing me? Two burn subscription quota I've already paid for, one spends real API dollars, and none showed up on the Grafana dashboards I'd built for everything else in the rack. So I fixed it. Every call from every AI in the lab — tokens, latency, cache hits, and dollars — now lands in Prometheus and one Grafana dashboard. This walks through the four measurement legs, the PromQL traps that made my first dashboard lie, and the privacy scrub that makes the screenshots publishable. Make the addresses your own. Every machine-specific value here is a placeholder: the monitoring host 10.0.0.5 , agent host 10.0.0.7 , Ollama nodes 10.0.0.1 – 10.0.0.3 , exporter ports, and any /home/youradmin paths. The AI layer has a genuinely weird cost structure. Two interactive CLIs run on flat subscriptions, so their "cost" is quota — a percentage of a weekly allowance. The automated agent calls a hosted API and pays per token. Same lab, three billing models. So the dashboard has two columns: quota burn a percentage that resets and real dollars the metered agent . The coding CLI needs no wrapper — it has native OpenTelemetry support https://code.claude.com/docs/en/monitoring-usage . Switch it on with env vars: export CLAUDE CODE ENABLE TELEMETRY=1 export OTEL METRICS EXPORTER=otlp export OTEL EXPORTER OTLP PROTOCOL=grpc export OTEL EXPORTER OTLP ENDPOINT=http://10.0.0.5:4317 Every session exports claude code.token.usage by type — input, output, cache read , claude code.cost.usage in USD, and claude code.session.count . Those arrive over OTLP, which Prometheus doesn't scrape directly — so an OpenTelemetry Collector https://opentelemetry.io/docs/collector/ on the monitoring host listens on :4317 , applies processors, and re-exposes everything in Prometheus format on :8889 . The automated agent doesn't speak OTel, but it writes an honest log line per call: provider, model, token counts, latency, cache hit, and the occasional "fallback activated". A ~200-line stdlib-only Python exporter tails the log with a regex, keeps counters in memory, and serves them on :9109 . It also reads the agent's task database read-only for queue-depth gauges — so the dashboard shows quality alongside spend: is the agent finishing work, or just burning tokens? My three-node Ollama cluster was stubborn: Ollama ships no Prometheus endpoint as of the 0.30 releases . But it leaks what you need in two places — its journald request log status, latency, caller IP and the documented /api/ps endpoint which models are loaded right now . One exporter per node on :9110 . The caller-IP label turned out useful: it shows who is using the local models. The second CLI Codex https://github.com/openai/codex records tokens and quota in session files, but the laptop sleeps and moves, so Prometheus can't reliably scrape it. That's exactly the Pushgateway https://prometheus.io/docs/practices/pushing/ case: a systemd user timer parses the session files every 5 minutes and pushes lifetime totals plus quota percentage to :9091 . Wiring it together is four scrape jobs: scrape configs: - job name: ai claude code static configs: { targets: "10.0.0.5:8889" } OTel Collector - job name: ai codex honor labels: true static configs: { targets: "10.0.0.5:9091" } Pushgateway - job name: ai agent static configs: { targets: "10.0.0.7:9109" } log-tail exporter - job name: ai ollama static configs: { targets: "10.0.0.1:9110", "10.0.0.2:9110", "10.0.0.3:9110" } Trap 1: per-session counters break increase . Claude Code's counters are per-session and ephemeral; a short session leaves one sample, and range functions need two, so increase returns nothing while a real 30k-token session sits invisible. Read the last value each session reported and sum: sum max over time claude code token usage tokens total 1d Trap 2: composite cost math collapses on absent series. In PromQL, arithmetic with an empty operand makes the whole expression empty — so before any cache reads existed, real spend rendered as $0.00. Guard every component with or vector 0 : sum rate input tokens 1h or vector 0 1.00 / 1e6 + sum rate cache tokens 1h or vector 0 0.10 / 1e6 + sum rate output tokens 1h or vector 0 5.00 / 1e6 Trap 3: histogram quantile returns literal NaN over idle windows. Documented behavior with zero observations — my latency panel drew garbage across every quiet hour. Consumers need to drop non-finite samples; Grafana panels just go sparse when the lab is idle, which is the honest picture. Claude Code's telemetry attaches identity by default — your email , account ids, org id — as labels on every metric. Useful in a company; radioactive on a public screenshot. Going forward, the OTel Collector deletes those before they reach Prometheus: processors: attributes/scrub: actions: - { key: user.email, action: delete } - { key: user.account uuid, action: delete } - { key: user.account id, action: delete } - { key: user.id, action: delete } - { key: organization.id, action: delete } Delete by key, not value, so anyone who ever exported from that laptop gets scrubbed. For history already on disk, open a temporary Prometheus admin window --web.enable-admin-api , delete the identity-labeled series, then close it: curl -X POST 'http://10.0.0.5:9090/api/v1/admin/tsdb/delete series?match ={user email =""}' curl -X POST 'http://10.0.0.5:9090/api/v1/admin/tsdb/clean tombstones' now remove --web.enable-admin-api from the unit and restart again One deliberate non-deletion: I kept session id . Dropping it merges per-session cumulative counters into one series last write wins, totals undercount . Scrub identity; keep cardinality that's structurally load-bearing. My Node Down alert was up == 0 . A sleeping laptop is not an outage, so it's now scoped to exclude the AI jobs — a dead AI exporter shows as a dashboard gap, real infra still pages: up{job ~"ai . "} == 0 Not spend — the cache-hit rate . At ~78% cached input, the agent's metered bill stays in coffee money. The day that rate drops is the day something changed in how it builds prompts, and now I'll see it the same morning. Total cost of the measurement layer: two tiny Python exporters, one collector, one gateway, and an evening. Originally published on peira.dev https://peira.dev/blog/ai-spend-telemetry-grafana/ .