Repository navigation
fix(codex): raise GPT-6.1 Sol generated context default to 500k - #8033
Merged
Merged
Conversation
Contributor
|
All contributors have signed the CLA. ✅ |
Contributor
Author
|
I have read the CLA Document and I hereby sign the CLA |
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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When Codex uses Sub2API's generated GPT-6.1 Sol catalog (including an API-key model endpoint that omits context metadata), it still receives the bundled 272,000-token default and may compact earlier than the newly announced Codex/Work default.
Set the generated default to 500,000 tokens. Keep the pinned official JSON snapshot intact, preserve explicit upstream context values, and leave the existing maximum-window and billing policies unchanged. Regression coverage checks the offline catalog, missing/null API-key context metadata, lower/higher explicit upstream windows, and byte-for-byte OAuth manifest passthrough.
Related: #8032. The change follows the October 12 announcement reported there; the checked-in openai/codex static catalog still advertises the older default. This is a generated-catalog default update, not evidence that every upstream account has received the rollout.
Validation (October 12, 2026, macOS arm64 / Go 1.27.2):
CI=true CGO_ENABLED=0 GOPROXY=https://goproxy.cn DOCKER_HOST=unix:///Users/xugao/.colima/default/docker.sock TESTCONTAINERS_DOCKER_SOCKET_OVERRIDE=/var/run/docker.sock go test -p 1 -tags=integration ./.... PostgreSQL 18.1 and Redis 8.4 test containers ran; Docker-unavailable skipping was prevented with CI=true.CGO_ENABLED=0 GOPROXY=https://goproxy.cn golangci-lint run --concurrency 2 --timeout 30m ./..., using CI's golangci-lint v2.14.0; 0 issues.CGO_ENABLED=0 GOPROXY=https://goproxy.cn go test -p 2 -tags=unit ./..., then the complete suite retried with-parallel 1. Both runs failed onlyTestInflightEstimate_AccountMappingNoDBAndBoundedMemoryat billing_inflight_reservation_test.go:464: heap growth 8,575,720 / 8,589,064 bytes exceeded the unchanged 8,388,608-byte assertion. All other test packages passed. The isolated test passed three consecutive runs on both the PR and parent revision. The complete parent-revision service package passed with-parallel 1(220.999s), so the failure is not established as pre-existing. A subsequent isolated full PR unit run (-p 1 -parallel 1) passed the memory assertion but failedTestShouldAutoPauseGrokAccountByQuota/retry_after_active; that test uses parallel subtests and a 30-second real-time expiry, which can be sensitive to forced test serialization. The final full retry restored default test parallelism (go test -p 1 -tags=unit ./...) and again failed only the unchanged memory assertion. Root cause remains unresolved; no assertion was weakened, no test excluded, and the complete unit suite is NOT marked passing.