Codex CLI Overview (Beta)

⚠️ The Codex support is experimental. Prefer the ccusage codex commands; the @ccusage/codex package is now a compatibility wrapper.

The ccusage codex commands reuse ccusage's responsive tables, pricing cache, and token accounting to analyze OpenAI Codex CLI session logs.

Installation & Launch

# Recommended
npx ccusage@latest codex --help
bunx ccusage codex --help

# Alternative package runners
pnpm dlx ccusage codex --help
pnpx ccusage codex --help

# Using deno (with security flags)
deno run -E -R=$HOME/.codex/ -S=homedir -N='raw.githubusercontent.com:443' npm:ccusage@latest codex --help

Compatibility package

npx @ccusage/codex@latest daily still works during the migration window, prints a deprecation warning, and forwards to ccusage codex daily.

Recommended: Shell Alias

If you want a shorter command, set up a shell alias:

# bash/zsh: alias ccu-codex='bunx ccusage codex'
# fish:     alias ccu-codex 'bunx ccusage codex'

# Then simply run:
ccu-codex daily
ccu-codex monthly --json

TIP

After adding the alias to your shell config file (.bashrc, .zshrc, or config.fish), restart your shell or run source on the config file to apply the changes.

Data Source

The CLI reads Codex session JSONL files located under CODEX_HOME (defaults to ~/.codex). Each file represents a single Codex CLI session and contains running token totals that the tool converts into per-day or per-month deltas.

What Gets Calculated

• Token deltas – Each event_msg with payload.type === "token_count" reports cumulative totals. The CLI subtracts the previous totals to recover per-turn token usage (input, cached input, output, reasoning, total).
• Per-model grouping – The turn_context metadata specifies the active model. We aggregate tokens per day/month and per model. Sessions lacking model metadata (seen in early September 2025 builds) are skipped.
• Pricing – Rates come from LiteLLM's pricing dataset via the shared LiteLLMPricingFetcher. Aliases such as gpt-5-codex map to canonical entries (gpt-5) so cost calculations remain accurate.
• Speed pricing – --speed auto is the default. It reads ${CODEX_HOME:-~/.codex}/config.toml and applies fast pricing when Codex has service_tier = "priority" or legacy service_tier = "fast" configured. Fast mode uses the model-specific LiteLLM multiplier when available and otherwise falls back to 2x pricing. Pass --speed fast or --speed standard to override config-based detection.
• Legacy fallback – Early September 2025 logs that never recorded turn_context metadata are still included; the CLI assumes gpt-5 for pricing so you can review the tokens even though the model tag is missing (the JSON output also marks these rows with "isFallback": true).
• Cost formula – Non-cached input uses the standard input price; cached input uses the cache-read price (falling back to the input price when missing); and output tokens are billed at the output price. All prices are per million tokens. Reasoning tokens may be shown for reference, but they are part of the output charge and are not billed separately.
• Totals and reports – Daily, monthly, and session commands display per-model breakdowns, overall totals, and optional JSON for automation.

Environment Variables

Variable	Description
CODEX_HOME	Override the root directory containing Codex session folders
LOG_LEVEL	Adjust log verbosity (0 silent … 5 trace)

When Codex emits a model alias (for example gpt-5-codex), the CLI automatically resolves it to the canonical LiteLLM pricing entry. No manual override is needed.

Speed Pricing

Codex logs usually do not include whether a turn used fast mode. By default, ccusage codex uses --speed auto, reads ${CODEX_HOME:-~/.codex}/config.toml, and treats service_tier = "priority" or legacy service_tier = "fast" as fast pricing. Fast mode uses the model-specific LiteLLM multiplier when available and otherwise falls back to 2x pricing.

# Default: read Codex config.toml
ccusage codex daily --speed auto

# Force fast pricing
ccusage codex daily --speed fast

# Force standard pricing
ccusage codex daily --speed standard

Next Steps

• Daily report command
• Monthly report command
• Session report command
• Additional reports will mirror the ccusage CLI as the Codex tooling stabilizes.

Have feedback or ideas? Open an issue so we can improve the beta.

Troubleshooting

Why are there no entries before September 2025?

OpenAI's Codex CLI started emitting token_count events in commit 0269096 (2025-09-06). Earlier session logs simply don't contain token usage metrics, so ccusage codex has nothing to aggregate. If you need historic data, rerun those sessions after that Codex update.

What if some September 2025 sessions still get skipped?

During the 2025-09 rollouts a few Codex builds emitted token_count events without the matching turn_context metadata, so the CLI could not determine which model generated the tokens. Those entries are ignored to avoid mispriced reports. If you encounter this, relaunch the Codex CLI to generate fresh logs—the current builds restore the missing metadata.