OpenAI ships GPT‑5.5

OpenAI released GPT‑5.5 on April 23 and added GPT‑5.5 and GPT‑5.5 Pro to the application programming interface a day later. (openai.com) The company described GPT‑5.5 as its “smartest and most intuitive to use model yet” and said it is built for coding, research, document work, spreadsheets, and other computer-based tasks. (openai.com) OpenAI said GPT‑5.5 is available in ChatGPT, Codex, and the application programming interface, with GPT‑5.5 Pro offered as a higher-compute setting on the same underlying model. (openai.com 1) (openai.com 2) A base model is the main engine underneath a chatbot, and test-time compute is the extra processing a model uses before answering. OpenAI said GPT‑5.5 Pro uses parallel test-time compute to spend more effort on difficult tasks. (openai.com) OpenAI said the new model improves on GPT‑5.4 in agentic coding, tool use, and multi-step work across documents, spreadsheets, and slide decks while keeping GPT‑5.4-level latency. (openai.com 1) (openai.com 2) The company’s system card said GPT‑5.5 is designed to understand tasks earlier, ask for less guidance, use tools more effectively, check its own work, and continue until a task is finished. (openai.com) OpenAI updated the system card on April 24 to add safeguards for the application programming interface rollout of GPT‑5.5 and GPT‑5.5 Pro. The same card said most safety results for GPT‑5.5 are treated as proxies for GPT‑5.5 Pro, with separate testing where extra compute could change risk. (openai.com) The release also settled a closely watched prediction market. Polymarket showed the “GPT‑5.5 released by...?” market at 100% after OpenAI’s April 23 announcement, with more than $1.1 million traded as of that date. (polymarket.com) The rollout arrived as OpenAI faced a separate safety controversy in Canada. CBS News reported on April 25 that Sam Altman said the company was “deeply sorry” for not alerting law enforcement after banning a ChatGPT account later tied to the Tumbler Ridge, British Columbia, school shooter. (cbsnews.com) Altman’s letter said the account had been banned in June 2025, about eight months before the shooting, according to CBS. That placed OpenAI’s newest model launch alongside new questions about how providers handle warning signs on their platforms. (cbsnews.com)