Skip to content

Validate publication cost provenance with a frozen offline contract - #172

Draft
MaxGhenis wants to merge 2 commits into
mainfrom
codex/issue118-cost-publication
Draft

Validate publication cost provenance with a frozen offline contract#172
MaxGhenis wants to merge 2 commits into
mainfrom
codex/issue118-cost-publication

Conversation

@MaxGhenis

@MaxGhenis MaxGhenis commented Sep 7, 2026

Copy link
Copy Markdown
Contributor

Retained prediction-row costs cannot establish complete billed spend or comparable synchronous estimates. This adds a frozen v1 cost-basis contract and an offline validator over the existing physical-call ledger, with separate receipts and dated normalization, exact model/call reconciliation, explicit unknowns, and synthetic fixtures.

The included frozen-release audit classifies 39 model rows across 100 scenarios and finds no call ledger in that bundle. It leaves all rows ineligible and marks missing attempts unrecoverable from the bundle. Public costs remain disabled in the new eligibility report; no public UI/export, serving treatment, reference output, launch-survivability, or stability code changes. Refs #118; this does not close it or satisfy its remaining release gates.

Validation on final head 9870b7c: all six CI checks pass, including 848 Python tests (8 skipped, 12 deselected), app, lint, and paper. The local requested ledger/cost-latency/report-cost/supervisor, Batch, and validator/audit suites pass (166 tests). Offline locked dependency sync and wheel/schema packaging pass. Both report CLIs ran with sockets denied: zero network attempts and no provider or PolicyEngine computation modules imported.

Independent coordinator review is pending. Source hashes and billing attribution remain attestations for that review. No real provider pricecard or model response was fetched. Keep this PR draft; public restoration, model reruns, provenance recovery, UI gates, and release approval remain separate work.

@vercel

vercel Bot commented Sep 7, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated
policybench-site Ready Ready Preview Sep 7, 2026 11:12pm UTC

Request Review

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant