GPT‑5.5 moves to workflows
- OpenAI said on April 23 that GPT‑5.5 is rolling out in ChatGPT and Codex, framing it as a model for coding, research, spreadsheets, and software tasks rather than chat alone. - The company said GPT‑5.5 matches GPT‑5.4’s serving latency, uses fewer tokens on Codex tasks, and is reaching Plus, Pro, Business, and Enterprise users, with API access added a day later. - The shift comes as OpenAI has already retired several older ChatGPT models, leaving API access in place and pushing teams toward model governance and fallback planning. (help.openai.com)
OpenAI launched GPT‑5.5 on April 23 and positioned it as a model for getting work done across software, documents, research, and coding tools, not just answering prompts. (openai.com) The company said GPT‑5.5 is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, while GPT‑5.5 Pro is going to Pro, Business, and Enterprise users in ChatGPT. OpenAI updated the post on April 24 to say GPT‑5.5 and GPT‑5.5 Pro are also available in the application programming interface. (openai.com) OpenAI described the model as better at planning, using tools, checking its own work, and carrying multi-step tasks through to completion. In Codex, OpenAI said it is the recommended model for implementation, refactors, debugging, testing, validation, and knowledge-work artifacts. (openai.com) (developers.openai.com) The product pitch marks a change in emphasis from chatbot fluency to workflow execution. OpenAI said GPT‑5.5 can research online, analyze data, create documents and spreadsheets, operate software, and move across tools until a task is finished. (openai.com) OpenAI also tied the release to infrastructure, saying GPT‑5.5 matches GPT‑5.4 per-token latency in real-world serving while delivering higher performance. The company said the model uses significantly fewer tokens to complete the same Codex tasks. (openai.com) (community.openai.com) On OpenAI’s published benchmarks, GPT‑5.5 scored 82.7% on Terminal-Bench 2.0 versus 75.1% for GPT‑5.4, and 78.7% on OSWorld-Verified versus 75.0% for GPT‑5.4. OpenAI also reported 84.4% on BrowseComp for GPT‑5.5 and 90.1% for GPT‑5.5 Pro. (openai.com) The rollout lands after OpenAI already cleared out older choices inside ChatGPT. Its help documentation says GPT‑4o, GPT‑4.1, GPT‑4.1 mini, OpenAI o4-mini, and GPT‑5 Instant and Thinking were retired from ChatGPT on February 13, 2026, while API access remained unchanged. (help.openai.com) That separation matters for companies building on the models. OpenAI’s help page says Business, Enterprise, and Edu customers retained GPT‑4o access inside Custom GPTs until April 3, 2026, even as the consumer-facing picker changed. (help.openai.com) Codex’s own changelog shows the same workflow push. On April 16, OpenAI said Codex was becoming “a broader workspace for getting work done with AI,” adding an in-app browser, computer use for macOS apps, and longer-running task features before GPT‑5.5 arrived a week later. (developers.openai.com) OpenAI’s system card says nearly 200 trusted early-access partners gave feedback before release, and the April 24 update added safeguards for serving GPT‑5.5 and GPT‑5.5 Pro in the application programming interface. (openai.com 1) (openai.com 2) The result is less a simple chatbot upgrade than a product stack built around model turnover, tool use, and long-running tasks. OpenAI is selling GPT‑5.5 as the default worker inside that stack, with older models increasingly treated as legacy options. (openai.com) (help.openai.com)
