OpenAI 从 Codex 移除 gpt-5.6-sol 的 Ultrafast 档，疑为 GPT-6 Sol 让路
爆料称 Codex 删掉了 gpt-5.6-sol 的 Ultrafast 服务档，而 API 里该档仍在，时间点临近 DevDay，被解读为 GPT-6 Sol 接位的信号。
Looks like OpenAI might be preparing to release Ultrafast for GPT-6 Sol, maybe this week or around DevDay 👀👀 Follow-up to my last post: OpenAI just removed the ultrafast service tier from gpt-5.6-sol in Codex, even though Ultrafast itself still exists in OpenAI’s API. The removed entry literally described it as: “The fastest available responses for latency-sensitive work.” Ultrafast can already push GPT-5.6 Sol up to 750 output tokens/sec / 14× Standard speed using Cerebras. Could just be product cleanup, but if GPT-6 Sol is about to replace 5.6 Sol’s spot, removing Ultrafast from the old Codex entry right before DevDay is pretty interesting. Codex change: https://github.com/openai/codex/pull/47130 OpenAI Ultrafast: https://openai.com/index/previewing-ultrafast/