v0.3.3 — GPT-6.1 Sol support

Adds GPT-6.1 Sol pricing: $2 input, $0.10 cached input, and $10 output per million tokens. Adds model display labels and updates the README pricing reference.

Codex Token Watch v0.3.2

Removes the API long-context multiplier from cost comparisons. API EST. now uses base Standard API token rates for every model call and remains a comparison, not a Codex subscription charge or an exact API bill. README wording has been updated accordingly.

Codex Token Watch v0.3.1

Adds GPT-6 Sol and GPT-6 Luna Standard API-equivalent pricing, model labels, and GPT-6 Luna support for --hide-luna. Estimates apply the documented long-context rates per model call above 272K input tokens. Source: https://developers.openai.com/api/docs/pricing

Codex Token Watch v0.3.0

What changed

• Add --hide-luna to hide Luna-only turn and cache-miss rows.
• Keep hidden Luna usage in ALL and summary totals.
• Keep mixed-model turns visible.
• Show an explicit count of hidden Luna rows so totals remain easy to interpret.
• Speed up --last N discovery by expanding through indexed recent-task windows instead of scanning every historical task record.

Default output remains unchanged when the flag is omitted.

Codex Token Watch v0.2.0

What changed

• Add codex-token-watch --last N for the latest turns across tasks.
• Add codex-token-watch --since 6h for every turn in a rolling time window.
• Show aggregate token, cache, output, reasoning, cost, call, and duration totals for the selected turns.
• Include task, model, effort, change, and prompt context in the cross-task report.
• Keep count and time-window modes separate; combining --last and --since returns a clear error.

The original task-specific and cache-miss commands remain unchanged.

Codex Token Watch v0.1.2

What changed

• Show the model and reasoning effort for each cross-task cache miss.
• Show model or effort transitions in a compact CHANGE column.
• Display — when the turn has no settings change.
• Preserve the responsive prompt layout introduced in v0.1.1.

The report reuses metadata from the existing rollout scan and adds no extra search pass.

Codex Token Watch v0.1.1

What changed

• Show the turn prompt in cross-task cache-miss reports.
• Use an inline PROMPT column when the terminal is wide enough.
• Fall back to a separate PROMPTS section on narrow terminals.

Prompt display reuses data already parsed during the existing scan, so it does not add another search pass.

Codex Token Watch v0.1.0

Initial public release of Codex Token Watch.

Highlights:

• Per-turn input, cache, output, reasoning, model-call, and duration reporting
• API-equivalent cost estimates
• Model and reasoning-effort change detection
• Complete cache-miss analysis for one task
• Cross-task cache-miss reports with rolling time windows
• Next-turn cache recovery percentages
• Local, read-only operation with no background process or third-party dependencies