Repository navigation
Conversation
A token_count event with info: null carries only rate limits, so it no longer marks the session's request collection as partial. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
CombineCosts flags partial only for partial members or unpriced tokens, InferenceGeo is lower-cased once for every model, and the unreachable gpt-daybreak-blue clause in requestContext is gone. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
RequestCost is the single pricing formula; the former CostUSD tests now exercise it through the aggregate-usage path. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A failed ListChildren call is logged and the parent sorts by its own cost instead of failing the whole page. Session rows keep a single cost summary rather than pricing twice. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
RequestProjectionVersion and LegacyProjectionVersion replace the bare 14 and 13 literals in the push validators. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Current model tariffs were missing from the catalog, and session-level token totals could apply the wrong context or cache rate. This change adds the missing public models across OpenAI, Anthropic, Google, xAI, Z.AI, and Cursor, and prices measured inference requests before aggregating session and report costs.
Request facts preserve the effective model, timestamp, prompt length, tier, speed, region, and cache-write duration. Projection 14 carries those facts through SQLite, Postgres, and sync; uploads with the same raw hash preserve newer projections. Historical tariff periods remain date-sensitive, and CLI/panel output distinguishes calculated, estimated, partial, and unavailable costs.
Validation: full race suite with real Postgres, CLI integration tests, build, vet, golangci-lint, generated-code reproducibility, tidy check, and repository quality gates.
Rollout: update the server before CLI/panel clients. The next sync reimports available source files; sessions whose sources are unavailable retain aggregate estimates.