Tag
OpenAI's Tibo (@thsottiaux) announced that the default context length for GPT-6.1 Sol has been increased to 500k tokens when used via Codex and Work, up from 272k (~80% increase), with plans to go higher to reduce compactions on context-heavy tasks.
OpenAI 宣布 Ultrafast 功能今日起面向 GPT-6.1 Sol 在 API、Codex 和 ChatGPT Work 中推出。该开发者账号线程还涵盖 GPT-Live-1 语音 API、Codex 键盘快捷支与 Git 集成改进、Open Responses 开源规范及 Codex Skills 支持。
Microsoft accidentally leaked that GPT-6.1 Sol reuses the same base weights as GPT-6 Sol but uses only 2 inference passes instead of 3, revealing architectural details about OpenAI's model iteration.
A user claims that OpenAI's GPT-6.1 Sol generates text about 2.5x slower than GPT-6 Sol and 2.3x slower than GPT-5.6 Sol, suggesting OpenAI is throttling compute for paying customers to reallocate resources to internal experiments.
OpenAI cut the usage allotment of its $200/month Pro subscription in half, but simultaneously released GPT 6.1 Sol — a flagship model priced 10x to 77x cheaper than competing frontier models — leading the author to argue AI model economics are improving, not collapsing.
OpenAI's GPT-6.1 Sol replaces GPT-6 Sol after just 7 days, scoring near GPT-6 Astra on the Intelligence Index while costing less than a quarter per task and pushing the cost-efficiency Pareto frontier.
OpenAI launched GPT-6.1 Sol, an AI model that nearly matches GPT-6 Astra's performance in agentic coding and professional tasks at a lower cost, with enhanced accuracy and safety features.