Repository navigation
v2.4.3
2.4.3 — Each call at its own price
Adoption, read 2026-10-11: downloads of @trazum/cli in the last 30 days: unavailable (fetch failed), GitHub stars: unavailable (HTTP 403), MCP registry latest 2.4.2.
A patch with no new command. It makes the one line work on a first run, and
prices every call as it was actually billed.
One line, nothing to configure
npx @trazum/cli billWith no path, bill reads Claude Code's own folder ($CLAUDE_CONFIG_DIR/projects,
~/.claude/projects, ~/.config/claude/projects) and names it. Sessions are
labelled by project, as from-claude-code has done since 1.77, so the bill says
which project spent the money. On a terminal it prints a statement: the total, a
table by project, a table by model, and what is not in the total. Piped, it still
prints the JSON receipt, so scripts see no change; --receipt and --table
force either. A folder of hundreds of transcripts is one line per shape.
Every call at the price of its own moment
A bill priced every call at the date the report ran. A DeepSeek log read during
peak hours came out entirely at peak, and read on a Saturday entirely off-peak;
a call inside a promotion lost it once the promotion ended. Each call is now
priced at its own timestamp. And a model's price can change inside a log:
DeepSeek cut V4 Flash's price at 04:00 UTC on 2026-09-10 and retired it the same
day, so the catalogue now carries both prices, each until the published instant.
Four providers re-read
- Google: every price matched; caching is automatic on every Gemini, the
minimums moved (4,096 on 3.6 Flash, 2,048 on 2.5 Flash), and the 2.5 models
read the cache at 10%, not the 25% this table carried. Gemini 3.8 Flash added. - DeepSeek: V4.1 Flash (
deepseek-flash) added at $0.15/$0.60; the old V4
Flash id bills at that price since the cut and at its own before it. - Mistral: every price matched, and it caches: cached input at 10%. How it
starts and its minimum are not published anywhere Trazum can read, so no
caching advice is given for it. - OpenAI: every price matched; the dated GPT-5 snapshot ids an OpenAI usage
export carries now price. The new generation (GPT-6, GPT-5.6) and Claude
Haiku 5.5 wait for per-request prompt sizes: each changes price above a prompt
length a usage log does not always state.
Elsewhere
- The web app opens on the bill, and its tour starts there.
@trazum/cli's npm listing names Claude Code, the sources it prices and the
PDF, so a search for what Claude Code cost can find it.