GPT‑5.5 ships with stronger autonomy and improved reasoning for agentic workflows
- OpenAI said on April 23 it began rolling out GPT‑5.5 to ChatGPT Plus, Pro, Business and Enterprise users, with GPT‑5.5 Pro for Pro, Business and Enterprise, plus Codex access. - OpenAI’s launch post said GPT‑5.5 scored 82.7% on Terminal‑Bench 2.0, up from GPT‑5.4’s 75.1%, and 78.7% on OSWorld‑Verified, while matching GPT‑5.4’s per‑token latency in serving. - OpenAI updated the release on April 24 to say GPT‑5.5 and GPT‑5.5 Pro were also available in the API after adding extra safeguards. (openai.com)
OpenAI says GPT‑5.5 is rolling out in ChatGPT and Codex as a model built for longer, messier work on a computer. (openai.com) In OpenAI’s April 23 release, GPT‑5.5 went to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, while GPT‑5.5 Pro went to Pro, Business, and Enterprise users in ChatGPT. (openai.com) (help.openai.com) OpenAI updated that post on April 24 to say GPT‑5.5 and GPT‑5.5 Pro were also available in the application programming interface, after adding extra safeguards for API deployment. (openai.com 1) (openai.com 2) The basic pitch is autonomy: a model that can take a multi-step task, plan the work, use tools, check results, and keep going without as much step-by-step prompting. OpenAI says that shows up in coding, online research, data analysis, spreadsheets, documents, and software use. (openai.com 1) (openai.com 2) OpenAI’s benchmark table shows GPT‑5.5 at 82.7% on Terminal‑Bench 2.0 versus 75.1% for GPT‑5.4, 84.9% on GDPval wins-or-ties versus 83.0%, and 78.7% on OSWorld‑Verified versus 75.0%. (openai.com) The company also says GPT‑5.5 matches GPT‑5.4 on per-token latency in real-world serving while using fewer tokens on the same Codex tasks. That frames the release as a capability increase without a speed penalty. (openai.com) For developers, OpenAI’s model docs now list `gpt-5.5` as the flagship model for complex reasoning and coding. The API pricing page lists it at $5 per 1 million input tokens and $30 per 1 million output tokens. (developers.openai.com) (openai.com) OpenAI’s Help Center says ChatGPT can route some “Instant” requests to GPT‑5.5 Thinking for harder jobs, while GPT‑5.5 Pro is positioned as the highest-capability option for long-running workflows. (help.openai.com) The safety side moved in parallel with the product launch. OpenAI says it ran full preparedness evaluations, added targeted red-teaming for cybersecurity and biology risks, and gathered feedback from nearly 200 early-access partners before release. (openai.com 1) (openai.com 2) That leaves GPT‑5.5 as OpenAI’s new top-line model for users who want an assistant to do more of the job itself, not just answer the next prompt. (openai.com) (developers.openai.com)
