Skip to content

v2.7.33-preview.20260722

Pre-release
Pre-release

Choose a tag to compare

Published to npm as @bitkyc08/opencodex@2.7.33-preview.20260722 with dist-tag preview.

Commit log

Changes since v2.7.31:

  • fix(responses): strip mismatched agent_message item ids before passthrough (#215) (90d1ef4)
  • fix(adapters): forward prompt_cache_key through openai-chat adapter (#217) (#224) (15d83ba)
  • fix(i18n): complete RU parity with dev en.ts + refresh RU docs (#211) (62998b4)
  • fix(update): fall back to direct proxy start when service reinstall fails (#227) (a1cdf49)
  • feat(google): add gemini-3.5-flash-lite to catalog (746f52b)
  • fix(codex): clamp catalog reasoning efforts to installed binary's ladder (b7ce5aa)
  • fix(usage): add qwen3.8-max-preview temporary Routeway pricing for both alibaba token plans (3387a9d)
  • fix(oauth): resolve Kiro login flow before blocking on manual paste (cec8c1a)
  • devlog: antigravity effort routing docs-first WP0 (000_plan + 001_research + 010_models + 020_adapter) (4930472)
  • antigravity: collapse effort-variant models into base IDs with effort routing infrastructure (c07f2d6)
  • antigravity: CCA buildRequest effort routing via resolveAntigravityEffortWireModel + thinkingConfig (140c0a8)
  • antigravity: effort routing fixtures + oauth reconcile test for collapsed model list (f3d2269)
  • devlog: archive antigravity effort routing to _fin (WP0-WP3 DONE) (33ee626)
  • antigravity: add claude-sonnet-4-6 effort control via thinkingConfig (Anthropic adaptive thinking verified) (03eecc4)
  • devlog: add sonnet 4.6 thinking verification to research doc (7859275)
  • models: add custom model config type (973e890)
  • models: add custom model management API (24775ad)
  • models: merge custom models into catalog (8f2922a)
  • models: fix custom model dedup + decodeURIComponent guard (aaf75bf)
  • feat(openrouter): add configurable provider routing (3c6f3ca)
  • gui: add custom model chip UI — header button, modal, hover popup, row badges (03dcbb9)
  • gui: fix tooltip flip-up boundary to match max-height (0152642)
  • i18n: add custom model chip keys for zh/de/ru (a5678df)
  • gui: enlarge model hover tooltip (1a52ebc)
  • cli: add models add/remove/list-custom subcommands (5c30a96)
  • cli: harden help — models subcommands + v2 entry (6d56810)
  • fix(server): sanitize reasoning content in v1 compact endpoint (df7661c)
  • fix(adapter): ensure root object type for Kimi tool schemas (d5b3796)
  • fix(antigravity): dedupe required tool fields (afae2f4)
  • cli: document v2 mode/threads subcommands in help (eb4841a)
  • feat(gui): add per-model cost breakdown to provider Usage tab (7ac3531)
  • devlog: remove archived antigravity effort routing originals (moved to _fin) (ef6d274)
  • fix(router): improve error message when no OpenAI provider is configured (353552f)
  • fix(cli): clarify ocx init prompts to indicate default provider selection (3c1fdfb)
  • fix: resolve unanswered auth and provider issues (2aa4430)
  • fix(combo): strip content-encoding on re-serialized child requests (#230) (afce7cf)
  • fix(gui): remove non-actionable API-key rows from Accounts tab (#231) (4d61025)
  • feat: add Japanese (ja) localization across GUI, docs-site, and README (#244) (a1fdbdc)
  • fix(config): replace stale pre-migration backup instead of crashing (#258) (c3517ce)
  • ci: improve issue intake, enforce minimum quality, and ping PR authors on wrong-branch retarget (#229) (07d7b91)
  • fix: stabilize merged PR integrations (fdacd14)
  • docs: document maintainers and review ownership (#264) (3a5f984)
  • fix(init): preserve valid pre-migration rollback backup instead of unconditional delete (a41170c)
  • fix(init): publish preserved rollback snapshot with no-replace copy (2976356)

Full diff