- Activated GPT-5.5 and GPT-5.5 Pro as first-attempt models on supported Codex runtimes while preserving deterministic fallback behavior for older runtimes and non-entitled accounts. (#435)
- Fixed unsupported-model retry handling so ChatGPT-backed Codex surfaces fall back to
gpt-5.4only after real upstream unsupported-model or no-access responses. (#435)
- Hardened request, auth, account, recovery, storage, settings, UI, and shutdown behavior around rate-limit backoff, token redaction, SSE buffering, account removal, manual-paste state, import size caps, Q hotkeys, OAuth state checks, retry writes, and cleanup idempotency. (#414, #417, #418, #419, #420, #421, #423, #424, #425, #426, #427, #428, #429, #431, #432)
- Removed dead command modules and ghost tests, refreshed hybrid-selector coverage for the null contract, and rolled open fix PRs into a single review branch before the GPT-5.5 activation. (#416, #422, #432)
- Validated the runtime rollout with build and focused model-map, transformer, wrapper, config, capability, fetch-helper, and plugin suites.