Add Daybreak model pricing

[0xSMW]

Summary

• price Daybreak Blue at GPT-5.6 Sol rates while preserving its runtime identity
• price Daybreak Red at published GPT-5.6 Cyber rates; leave undocumented dimensions unavailable
• require a complete long-context catalog tuple for models with explicit-only pricing
• preserve models.dev overrides and invalidate cached estimates when bundled pricing changes

Proof

• exact head: 7b593f52e8e4ab35080ff2f43dadaa3cc8a59c6f
• redacted real local-session aggregate recorded gpt-daybreak-blue-latest
• cumulative counters: 5,398,304 input, 4,823,366 cached input, 57,521 output, 5,455,825 total
• computed standard cost: (574,938 × $5 + 4,823,366 × $0.50 + 57,521 × $30) / 1M = $7.012003
• counters are cumulative session/thread totals; this evidence does not claim independent per-call attribution
• focused pricing suite: 57 tests passed
• architecture gatekeeper suite: 38 tests passed
• full test suite: 856 selections across 72 groups passed with 0 retries, failures, or timeouts
• make check: passed
• GitHub CI: 9/9 checks passed on the exact head

Sources: Blue model, Red model, and API pricing.

[clawsweeper]

🦞👀
ClawSweeper picked this up.

Pull request received. I will update this pull request when review starts.

ClawSweeper review complete

ClawSweeper finished reviewing this revision. The review result is being finalized.

View the workflow run.

[clawsweeper]

Codex review: blocked before merge. Reviewed September 5, 2026, 10:55 AM ET / 14:55 UTC.

ClawSweeper review

What this changes

Adds Daybreak cost estimates, preserves local model identities and pricing overrides, rejects undocumented pricing dimensions, and updates regression tests and cache invalidation metadata.

Merge readiness

⛔ Blocked before merge - 3 items remain

Keep open: the bundled Daybreak support remains absent from main and v0.56.6. Previous findings are addressed, and this review found no new blocking source defect.

Priority: P2
Reviewed head: b0577dda544083de37f529988929f09136d959ef

Review scores

Measure	Result	What it means
Overall readiness	🐚 platinum hermit (4/6)	A focused pricing extension with real-session aggregate evidence, targeted regression coverage, and no blocking source finding.
Proof confidence	🐚 platinum hermit (4/6) ✨ media proof bonus	Sufficient (screenshot): The reported network-denied scan of a real local Codex session, inspected terminal image, and captured transcript connect the new Blue pricing to a $7.012003 aggregate estimate; they do not claim per-call attribution, while focused tests supplement the pricing-boundary evidence.
Patch quality	🐚 platinum hermit (4/6)	No actionable review findings were identified.

Verification

Check	Result	Evidence
Real behavior	Verified	Sufficient (screenshot): The reported network-denied scan of a real local Codex session, inspected terminal image, and captured transcript connect the new Blue pricing to a $7.012003 aggregate estimate; they do not claim per-call attribution, while focused tests supplement the pricing-boundary evidence.
Evidence reviewed	10 items	Applicable repository policy: Read the complete root policy and checked for nested policies and maintainer notes; none were found for the changed paths. Applied its pricing-provider isolation, model-name, focused-test, and Keychain-safety guidance. Verified introduced scope: Inspected the complete local base-to-head diff, including portions truncated in the supplied patches: four files, 397 additions and 14 deletions. Production changes are confined to pricing and its generated parser hash. Prior findings remain resolved: The current code normalizes Blue before excluding API Fast pricing, requires all four long-context dimensions for explicit-only models, checks the resolved threshold, and validates cache writes in the active lane. The added tests cover these cases. The latest completed review examined this same head and reported no findings.
Findings	None	None.
Security	None	None.

How this fits together

CodexBar reads local session token counts and combines them with custom, catalog, or bundled prices to estimate usage costs. This change extends the shared pricing layer used by its session scanners.

flowchart TD
  A[Local session token counts] --> B[Resolve model identity]
  B --> C[Custom pricing override]
  B --> D[Catalog then bundled prices]
  C --> E[Validate applicable rates]
  D --> E
  E --> F[Cost estimate or unavailable]
  F --> G[Usage reports]

Loading

Before merge

• Resolve merge risk (P1) - GitHub reports merge conflicts, so behavior after integration with current main has not been established.
• Resolve merge risk (P1) - Current official Daybreak alias and rate documentation could not be independently retrieved because DNS failed.
• Complete next step (P2) - Resolve conflicts against current main, preserve its newer pricing behavior, regenerate the parser hash and gatekeeper coordinates, and validate the refreshed head with focused pricing tests, make test, and make check.

Agent review details

Security

None.

Review metrics

Metric	Value	Why it matters
Production and test delta	Production +95/−6; tests +302/−8 across four files	Production growth implements model-specific pricing availability; nine added tests cover aliases, catalog precedence, and context boundaries.

Merge-risk options

Maintainer options:

1. Decide the mitigation before merge
   Retain one shared pricing resolver with documented Daybreak fallbacks, exact overrides, unavailable unsupported dimensions, and automatic repricing of cached usage.
2. Pause or close
   Do not merge this PR until maintainers decide whether the risk is worth taking.

Technical review

Best possible solution:

Retain one shared pricing resolver with documented Daybreak fallbacks, exact overrides, unavailable unsupported dimensions, and automatic repricing of cached usage.

Do we have a high-confidence way to reproduce the issue?

Not applicable as a bug reproduction: this adds bundled model support. The contributor supplies a real-session Blue example, and source inspection confirms the missing entries on current main.

Is this the best way to solve the issue?

Yes: extending the existing shared fallback table preserves established override precedence and avoids a parallel pricing mechanism.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 1696c7a71c94.

Labels

Label changes:

• add rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🐚 platinum hermit and patch quality is 🐚 platinum hermit.
• remove rating: 🦞 diamond lobster: Current PR rating is rating: 🐚 platinum hermit, so this older rating label is no longer current.
• remove merge-risk: 🚨 other: Current PR review selected no merge-risk labels.

Label justifications:

• P2: This is a bounded improvement to local usage estimates for observed model identities.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🐚 platinum hermit and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (screenshot): The reported network-denied scan of a real local Codex session, inspected terminal image, and captured transcript connect the new Blue pricing to a $7.012003 aggregate estimate; they do not claim per-call attribution, while focused tests supplement the pricing-boundary evidence.
• proof: sufficient: Contributor real behavior proof is sufficient. The reported network-denied scan of a real local Codex session, inspected terminal image, and captured transcript connect the new Blue pricing to a $7.012003 aggregate estimate; they do not claim per-call attribution, while focused tests supplement the pricing-boundary evidence.
• proof: 📸 screenshot: Contributor real behavior proof includes screenshot evidence. The reported network-denied scan of a real local Codex session, inspected terminal image, and captured transcript connect the new Blue pricing to a $7.012003 aggregate estimate; they do not claim per-call attribution, while focused tests supplement the pricing-boundary evidence.

Evidence

What I checked:

• Applicable repository policy: Read the complete root policy and checked for nested policies and maintainer notes; none were found for the changed paths. Applied its pricing-provider isolation, model-name, focused-test, and Keychain-safety guidance. (AGENTS.md:1, b0577dda5440)
• Verified introduced scope: Inspected the complete local base-to-head diff, including portions truncated in the supplied patches: four files, 397 additions and 14 deletions. Production changes are confined to pricing and its generated parser hash. (Sources/CodexBarCore/Vendored/CostUsage/CostUsagePricing.swift:202, b0577dda5440)
• Prior findings remain resolved: The current code normalizes Blue before excluding API Fast pricing, requires all four long-context dimensions for explicit-only models, checks the resolved threshold, and validates cache writes in the active lane. The added tests cover these cases. The latest completed review examined this same head and reported no findings. (Sources/CodexBarCore/Vendored/CostUsage/CostUsagePricing.swift:673, b0577dda5440)
• Still distinct from current main: Current main contains the existing model pricing table and generic override mechanisms but no Daybreak or Cyber entries in source, tests, documentation, or changelog. Direct inspection of its resolver confirmed the missing bundled support; GitHub canonical search was unavailable because api.github.com could not be reached. (Sources/CodexBarCore/Vendored/CostUsage/CostUsagePricing.swift:193, 1696c7a71c94)
• Latest release boundary: The local v0.56.6 tag points at the inspected current-main commit, matching the supplied latest-release metadata; that release therefore lacks these bundled entries. (1696c7a71c94)
• Existing pricing and upgrade paths: The existing overlay remains ahead of catalog and bundled pricing. Native and Pi scanners consume the pricing fingerprint, so changed bundled rates trigger recalculation. A read-only hash calculation also matched the generated value b9a05c82f76634f5. No stored preferences or serialized pricing schema are changed. (Sources/CodexBarCore/Vendored/CostUsage/CostUsagePricingKey.swift:18, b0577dda5440)

Likely related people:

• steipete: Suggested for follow-up; no historical authorship or introduction is verified. (role: unverified routing candidate; confidence: low)
• Yuxin-Qiao: Suggested for follow-up; no historical authorship or introduction is verified. (role: unverified routing candidate; confidence: low)

Rating scale

Score	Internal tier	Crab rank	Meaning
6/6	S	🦀 challenger crab	Exceptional readiness
5/6	A	🦞 diamond lobster	Very strong readiness
4/6	B	🐚 platinum hermit	Good normal PR; ordinary maintainer review
3/6	C	🦐 gold shrimp	Useful, but confidence is limited
2/6	D	🦪 silver shellfish	Proof or implementation needs work
1/6	F	🧂 unranked krab	Not merge-ready
N/A	NA	🌊 off-meta tidepool	Rating does not apply

Overall follows the weaker of proof and patch quality.
Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

Workflow

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

History

Review history (9 earlier review cycles; latest 8 shown)

• reviewed 2026-08-13T09:22:13.817Z sha d7b1d50 :: needs real behavior proof before merge. :: none
• reviewed 2026-08-13T11:08:51.604Z sha d7b1d50 :: needs real behavior proof before merge. :: none
• reviewed 2026-08-14T06:25:27.900Z sha dbb5ccc :: needs changes before merge. :: [P2] Reject incomplete long-context pricing tuples
• reviewed 2026-08-14T07:27:13.202Z sha 088d3aa :: needs changes before merge. :: [P2] Validate cache-write availability in the active pricing lane
• reviewed 2026-08-14T07:51:34.525Z sha 7b593f5 :: needs maintainer review before merge. :: none
• reviewed 2026-08-14T08:32:32.633Z sha 7b593f5 :: needs maintainer review before merge. :: none
• reviewed 2026-08-18T05:56:39.451Z sha 497ba47 :: needs maintainer review before merge. :: none
• reviewed 2026-08-19T14:25:58.909Z sha b0577dd :: needs maintainer review before merge. :: none

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 31c6db31da

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[0xSMW]

Addressed the review findings in d7b1d50:

• provider-qualified local Blue IDs no longer receive API Fast pricing
• missing long-context prices become unavailable at the resolved catalog threshold
• public Blue and Red aliases are documented by the linked OpenAI model pages
• an isolated, network-denied scan of a real local session identified gpt-daybreak-blue-latest and priced 5,455,825 tokens at $7.012003

Focused pricing tests (55), architecture gatekeeper tests (38), and make check pass.

@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[0xSMW]

@clawsweeper re-review last review seems stuck

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[0xSMW]

Exact-head real-session proof for dbb5ccc869de1889a2eee156fd464ca46d7acf10:

source: redacted aggregate from one real local Codex session log
session date: 2026-08-12
network: not used
credentials / account IDs / session IDs / paths / prompts: REDACTED

model context:          gpt-daybreak-blue-latest
input_tokens:           5,398,304
cached_input_tokens:    4,823,366
output_tokens:             57,521
reasoning_output:          18,636 (included in output_tokens)
total_tokens:           5,455,825

uncached_input = 5,398,304 - 4,823,366 = 574,938

(574,938 × $5.00
 + 4,823,366 × $0.50
 + 57,521 × $30.00) / 1,000,000

RESULT: $7.012003

The exact cumulative total_tokens value occurs in one local log whose model context is gpt-daybreak-blue-latest. These are cumulative session/thread counters; this evidence does not claim independent per-call attribution.

Verification on the merged exact head:

• CostUsagePricingTests: 55 passed
• make test: 856 selections across 72 groups passed; 0 retries, failures, or timeouts
• make check: passed

@clawsweeper re-review
@codex review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dbb5ccc869

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[0xSMW]

The partial long-context pricing finding is fixed in 088d3aa98, its review thread is resolved, and the exact head passes focused pricing tests, architecture gatekeeper tests, make check, and the full 72-group suite with zero failures, retries, or timeouts. The redacted real-session proof remains attached above.

@codex review
@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 088d3aa98f

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[0xSMW]

The active-lane cache-write finding is fixed in 7b593f52e, its thread is resolved, and the exact head passes 57 focused pricing tests, 38 architecture gatekeeper tests, make check, and the full 72-group suite with zero failures, retries, or timeouts. The redacted real-session proof remains attached in the PR body.

@codex review
@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. What shall we delve into next?

Reviewed commit: 7b593f52e8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[0xSMW]

Final exact-head verification is complete: all 9/9 GitHub checks pass on 7b593f52e, automated review completed with +1, all review threads are resolved, local focused suites and make check pass, the full local 72-group suite has zero failures/retries/timeouts, and the accepted redacted behavior proof is attached.

@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[0xSMW]

@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

[0xSMW]

@clawsweeper re-review

[clawsweeper]

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event exact_review_queue).
Result: when the review finishes, ClawSweeper will create the durable review comment if needed or update the existing comment in place.

Re-review progress:

• State: Complete
• Detail: The durable review result and its route handoff completed.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/32264574531
• Updated: 2026-08-19T14:32:36.287Z

[steipete]

Thanks @0xSMW for the pricing work and tests. The proposed IDs and aliases could not be corroborated in either the repository’s models.dev fixture or the public catalog checked September 22. This does not establish that the IDs are invalid, but leaves the bundled mappings unverified. #3676 (5e15f7e4f060) supplies provider-scoped catalog pricing; it did not land these aliases. Closing this change until the exact IDs and applicable rate dimensions are independently established.