OpenAI folds Codex into GPT-5.5

OpenAI released GPT-5.5 on April 23 and made it the recommended model in Codex, folding its coding push into one flagship model. (openai.com 1) (openai.com 2) Codex is still the product name for OpenAI’s coding agent, app, command-line tool, and integrated development environment extension. The dedicated model names under it have changed fast: GPT-5.2-Codex arrived in December 2025, GPT-5.3-Codex in February 2026, and GPT-5-Codex in September 2025 before GPT-5.5 took over this week. (openai.com 1) (openai.com 2) OpenAI’s April 23 Codex changelog says GPT-5.5 is now available in Codex and is the “recommended choice for most Codex tasks.” Users can select it in the Codex app, the command-line interface, and the integrated development environment extension as the rollout reaches accounts. (openai.com) Coding models write, edit, test, and debug software; “agentic” coding means the model can keep working through a task by using tools, checking results, and making follow-up changes instead of stopping after one answer. OpenAI says GPT-5.5 is built to do that across terminals, browsers, documents, spreadsheets, and other software. (openai.com 1) (openai.com 2) OpenAI says GPT-5.5 “uses significantly fewer tokens” to finish the same Codex tasks, which cuts the amount of text processed and can lower cost. The company also says it matched GPT-5.4’s per-token latency while improving coding and tool-use performance. (openai.com) In OpenAI’s published benchmarks, GPT-5.5 scored 82.7% on Terminal-Bench 2.0 versus 75.1% for GPT-5.4, and 73.1% on its internal Expert-SWE evaluation versus 68.5% for GPT-5.4. Those are company-run measurements, not independent audits, but they are the numbers OpenAI is using to sell the shift. (openai.com) The company is also widening where the model shows up. OpenAI says GPT-5.5 rolled out first to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex on April 23, and added GPT-5.5 and GPT-5.5 Pro to the application programming interface on April 24. (openai.com) That leaves Codex looking less like a separate family of models and more like a software layer on top of OpenAI’s main models. OpenAI’s September 2025 Codex update described GPT-5-Codex as “a version of GPT-5 further optimized for agentic coding”; this week’s rollout instead centers the base GPT-5.5 name inside Codex itself. (openai.com) (openai.com) OpenAI’s pitch is that one model can now handle code, research, data analysis, documents, and computer use without forcing customers to pick a separate coding brain first. The practical change for developers is simpler model selection: in Codex, GPT-5.5 is now the default place to start. (openai.com) (openai.com)