coding-agent-cost 0.1.1
Published to PyPI.
- Improve the README introduction, usage guidance and reproducible synthetic output example (#7). The historical example remains pinned to published 0.1.0.
- Add Codex token rates for exact model ID
gpt-6-astra(#8), with Standard/Fast estimates from confirmeddefault/priorityrequest settings and matching owned turn/model context. - Preserve tokens as
unpricedwhen attribution is missing or contradictory instead of silently pricing uncertain Astra usage as Standard. - The catalog observation cutoff is 2026-09-05T17:18:23Z (2026-09-06 02:18:23 JST). Earlier Astra usage remains unpriced. This is not an official launch or tariff commencement time.
- Target format and limitations:
docs/astra-pricing.md, based on public Codex 0.153.4 sources. These are request-setting estimates, not confirmation of the actual processing tier or bill. API-specific conditions and legacy message billing are outside this tariff. - Known GPT-5.6-family price discrepancies remain uncorrected. Missing historical events, alternative storage formats and ambiguous model/mode changes are not claimed as fully supported.
Source and verification
Build commit: c520df2368473579c58145c6d91e2bbded1ba40c.
Tag v0.1.1 points to merged commit 48c2731c3336401470d9d9016a98f437390abeef. Both commits have the same Git tree (743f27e4a05b7019c0d24674c003bb3aa4273cab); the retained artifacts were published without rebuilding.
Before publication, 177 source tests and 177 installed-wheel tests passed, as did twine check and catalog validation. After publication, both PyPI artifact SHA-256 values were verified. A fresh isolated installation from production PyPI without cache confirmed CLI/package/module version 0.1.1, import from the installed package, and catalog validation. Synthetic report and measure/v1 results match prepublication: Standard 6,000,000 tokens / $122 / 3,050 credits; Fast 6,000,000 tokens / $305 / 7,625 credits; unknown retains all 6,000,000 tokens as unpriced (numeric zero does not mean free). The Claude synthetic example remains 2,000,500 tokens / $11.25 / 500 unpriced tokens. These are synthetic fixture checks, not real-log validation. No runtime dependencies were added.
Distribution SHA-256
coding_agent_cost-0.1.1.tar.gz:047dc5cfe73f0cb9fcb11948d65618b2e385e066c7907ab822154d922a76d249coding_agent_cost-0.1.1-py3-none-any.whl:5ac5f215285694f7177aabfa5f3cfd0bfaa21900135b97de7be2f7f22d131eb0