GPT-6.1 Sol lands within one index point of GPT-6 Astra at one fifth the token price

Sources: pricing, announcement, benchmark, benchmark, HN discussion
Summary: OpenAI's own pricing table lists GPT-6.1 Sol at about one fifth the per-token price of GPT-6 Astra, and Artificial Analysis places the two models within one index point of each other. The independent figures describe the maximum-effort variant of each model, and that measurement reports 267.64 seconds to first answer token for Sol against 305.56 seconds for Astra. Both halves of the launch claim hold against sources outside OpenAI's announcement.
Why it matters: Teams paying Astra rates for reasoning work can cut per-token cost about fivefold for a one-point index difference, and both models are batch and background tools rather than interactive ones at 267.64 and 305.56 seconds to first answer token.