Codex drops the ultrafast service tier from gpt-5.6-sol
GitHub·low signal
PR #47130, merged 2026-09-21, removes ultrafast from the bundled gpt-5.6-sol model's service_tiers, leaving priority (Fast) as its only advertised tier, and updates the multi-agent tool definition snapshots. A related commit in the same alpha window updates model catalog descriptions and gpt-5.6-sol priority. If you pinned a service tier in a Codex config or a multi-agent tool definition, that value no longer resolves for this model.