TREE NEWS update: OpenAI Codex product lead Tibo said the team has identified and fixed several quality defects affecting Astra users, with a reset scheduled before local midnight, or noon Beijing time. The issues included skills written for earlier models triggering too often or blocking the model from checking its own work, and an optional context-management experiment that was disabled after causing premature stops or replies to old messages. Roughly 4,000 to 5,000 users were affected by the experiment.
OpenAI Codex Lead Says Astra Quality Issues Fixed, Reset Coming by Midnight
AI take
A public admission that an optional context-management experiment degraded outputs for thousands of users is a reminder that agentic coding tools are still shipping features faster than they can be validated at scale. The specific failure modes — skills misfiring and the model being blocked from self-checking — point to how tightly orchestration layers, not raw model quality, now determine reliability. Whether the reset resolves these issues, and whether such experiments are gated more carefully before reaching users, is the open question.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.
Related News
25m ago
China Mobile’s AIDC Signed Capacity Tops 7GW as It Builds AI Data Centers
25m ago
China Mobile Launches AI Trusted Computing at 2026 Computing Power Conference
1h ago
China Accelerates Rollout of AI Compute Token Loans
1h ago
Xiongan International Computing Power Scheduling Center Begins Operation
2h ago
China’s national supercomputing internet tops 1.7M users
2h ago
BIS Warns AI Cuts Bank Patch Window From Weeks to Minutes