{"id":1294,"date":"2026-04-24T19:37:56","date_gmt":"2026-04-24T19:37:56","guid":{"rendered":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/"},"modified":"2026-04-24T19:37:56","modified_gmt":"2026-04-24T19:37:56","slug":"chatgpt-5-5-and-codex-feel-like-a-real-step-up","status":"publish","type":"post","link":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/","title":{"rendered":"ChatGPT 5.5 and Codex Feel Like a Real Step Up"},"content":{"rendered":"<p>I have been using ChatGPT 5.5 and Codex today, and the difference from 5.4 is immediately noticeable. This is not a careful benchmark, so I do not want to pretend it is more scientific than it is. But for code and logic execution, it feels like a real step up: fewer dead ends, better follow-through, and a stronger sense that the model understands the shape of the task rather than just the next patch.<\/p>\n<p>That lines up with how OpenAI is positioning the release. In its <a href=\"https:\/\/openai.com\/index\/introducing-gpt-5-5\/\">GPT-5.5 announcement<\/a>, OpenAI says the model is strongest in agentic coding, computer use, knowledge work, and research-style workflows. The <a href=\"https:\/\/developers.openai.com\/codex\/concepts\/subagents#model-choice\">Codex developer docs<\/a> are even more direct about Codex-style work: GPT-5.5 is described as the starting point for demanding agents when available, especially ambiguous multi-step work that needs planning, tool use, validation, and follow-through across a larger context.<\/p>\n<p>The bit that matters to me is not that it can write more code. Models have been able to produce a lot of code for a while. The improvement is that GPT-5.5 seems better at keeping the whole job in its head. It is more willing to inspect the surrounding system, make a plan, edit the right files, run the checks, notice when something did not line up, and keep going. GPT-5.4 was already useful, but it still had more of that stop-start feeling where you had to keep dragging it back to the actual goal.<\/p>\n<p>The NVIDIA write-up is probably the most interesting external example because it is less about a demo and more about operational use. NVIDIA says <a href=\"https:\/\/blogs.nvidia.com\/blog\/openai-codex-gpt-5-5-ai-agents\/\">more than 10,000 employees got early access to GPT-5.5-powered Codex<\/a>, with engineers using it in complex multi-file codebases and seeing debugging cycles move from days to hours. I would still treat that as launch-post language, but the shape of the claim matches what I saw today: the gain is in persistence and coordination, not just clever snippets.<\/p>\n<p>There is a tradeoff. GPT-5.5 is not just &#8220;5.4 but free and faster&#8221;. The <a href=\"https:\/\/developers.openai.com\/codex\/pricing#what-are-the-usage-limits-for-my-plan\">Codex usage docs<\/a> say it uses fewer tokens to reach comparable results, but the limits and availability are different, and API access is still its own rollout story. For day-to-day development, though, I care more about wasted loops than raw message counts. A model that takes slightly more careful work per request but gets through the task with fewer retries can be cheaper in the way that actually matters: less time spent supervising nonsense.<\/p>\n<p>My practical take is simple: if GPT-5.5 is available in your Codex picker, I would use it for the hard tasks first. Give it the messy refactor, the confusing bug, the test failure that needs actual reasoning, or the bit of product logic where a shallow answer is worse than no answer. Keep 5.4 or the smaller models around for routine work, but 5.5 already feels like the model I would reach for when I want Codex to carry a thread properly.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>After using GPT-5.5 in Codex, the improvement over GPT-5.4 feels less like a benchmark bump and more like better follow-through on real coding work.<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[26,13],"tags":[143,39,116,142,141],"class_list":["post-1294","post","type-post","status-publish","format-standard","hentry","category-ai","category-programming","tag-ai-coding","tag-chatgpt","tag-codex","tag-gpt-5-5","tag-openai"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/\" \/>\n<meta property=\"og:locale\" content=\"en_GB\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one\" \/>\n<meta property=\"og:description\" content=\"After using GPT-5.5 in Codex, the improvement over GPT-5.4 feels less like a benchmark bump and more like better follow-through on real coding work.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/\" \/>\n<meta property=\"og:site_name\" content=\"wade.one\" \/>\n<meta property=\"article:published_time\" content=\"2026-04-24T19:37:56+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/Wade-Logo-cropped.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1518\" \/>\n\t<meta property=\"og:image:height\" content=\"1506\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@wadewomersley\" \/>\n<meta name=\"twitter:site\" content=\"@wadewomersley\" \/>\n<meta name=\"twitter:label1\" content=\"Estimated reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"2 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/\"},\"author\":{\"name\":\"\",\"@id\":\"\"},\"headline\":\"ChatGPT 5.5 and Codex Feel Like a Real Step Up\",\"datePublished\":\"2026-04-24T19:37:56+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/\"},\"wordCount\":494,\"publisher\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/#\\\/schema\\\/person\\\/8b4739f8f8bb2cff5d792d4b8779fcc3\"},\"keywords\":[\"ai-coding\",\"chatgpt\",\"codex\",\"gpt-5-5\",\"openai\"],\"articleSection\":[\"AI\",\"Programming\"],\"inLanguage\":\"en-GB\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/\",\"url\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/\",\"name\":\"ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/#website\"},\"datePublished\":\"2026-04-24T19:37:56+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/#breadcrumb\"},\"inLanguage\":\"en-GB\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/2026\\\/04\\\/24\\\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/wade.one\\\/blog\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"ChatGPT 5.5 and Codex Feel Like a Real Step Up\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/#website\",\"url\":\"https:\\\/\\\/wade.one\\\/blog\\\/\",\"name\":\"wade.one\",\"description\":\"wade womersley - york based software engineer\",\"publisher\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/#\\\/schema\\\/person\\\/8b4739f8f8bb2cff5d792d4b8779fcc3\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/wade.one\\\/blog\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-GB\"},{\"@type\":[\"Person\",\"Organization\"],\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/#\\\/schema\\\/person\\\/8b4739f8f8bb2cff5d792d4b8779fcc3\",\"name\":\"Wade Womersley\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-GB\",\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/wp-content\\\/uploads\\\/2015\\\/02\\\/200px.png\",\"url\":\"https:\\\/\\\/wade.one\\\/blog\\\/wp-content\\\/uploads\\\/2015\\\/02\\\/200px.png\",\"contentUrl\":\"https:\\\/\\\/wade.one\\\/blog\\\/wp-content\\\/uploads\\\/2015\\\/02\\\/200px.png\",\"width\":202,\"height\":200,\"caption\":\"Wade Womersley\"},\"logo\":{\"@id\":\"https:\\\/\\\/wade.one\\\/blog\\\/wp-content\\\/uploads\\\/2015\\\/02\\\/200px.png\"}}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/","og_locale":"en_GB","og_type":"article","og_title":"ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one","og_description":"After using GPT-5.5 in Codex, the improvement over GPT-5.4 feels less like a benchmark bump and more like better follow-through on real coding work.","og_url":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/","og_site_name":"wade.one","article_published_time":"2026-04-24T19:37:56+00:00","og_image":[{"width":1518,"height":1506,"url":"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/Wade-Logo-cropped.png","type":"image\/png"}],"twitter_card":"summary_large_image","twitter_creator":"@wadewomersley","twitter_site":"@wadewomersley","twitter_misc":{"Estimated reading time":"2 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/#article","isPartOf":{"@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/"},"author":{"name":"","@id":""},"headline":"ChatGPT 5.5 and Codex Feel Like a Real Step Up","datePublished":"2026-04-24T19:37:56+00:00","mainEntityOfPage":{"@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/"},"wordCount":494,"publisher":{"@id":"https:\/\/wade.one\/blog\/#\/schema\/person\/8b4739f8f8bb2cff5d792d4b8779fcc3"},"keywords":["ai-coding","chatgpt","codex","gpt-5-5","openai"],"articleSection":["AI","Programming"],"inLanguage":"en-GB"},{"@type":"WebPage","@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/","url":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/","name":"ChatGPT 5.5 and Codex Feel Like a Real Step Up - wade.one","isPartOf":{"@id":"https:\/\/wade.one\/blog\/#website"},"datePublished":"2026-04-24T19:37:56+00:00","breadcrumb":{"@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/#breadcrumb"},"inLanguage":"en-GB","potentialAction":[{"@type":"ReadAction","target":["https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/wade.one\/blog\/2026\/04\/24\/chatgpt-5-5-and-codex-feel-like-a-real-step-up\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/wade.one\/blog\/"},{"@type":"ListItem","position":2,"name":"ChatGPT 5.5 and Codex Feel Like a Real Step Up"}]},{"@type":"WebSite","@id":"https:\/\/wade.one\/blog\/#website","url":"https:\/\/wade.one\/blog\/","name":"wade.one","description":"wade womersley - york based software engineer","publisher":{"@id":"https:\/\/wade.one\/blog\/#\/schema\/person\/8b4739f8f8bb2cff5d792d4b8779fcc3"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/wade.one\/blog\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-GB"},{"@type":["Person","Organization"],"@id":"https:\/\/wade.one\/blog\/#\/schema\/person\/8b4739f8f8bb2cff5d792d4b8779fcc3","name":"Wade Womersley","image":{"@type":"ImageObject","inLanguage":"en-GB","@id":"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/200px.png","url":"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/200px.png","contentUrl":"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/200px.png","width":202,"height":200,"caption":"Wade Womersley"},"logo":{"@id":"https:\/\/wade.one\/blog\/wp-content\/uploads\/2015\/02\/200px.png"}}]}},"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"jetpack-related-posts":[{"id":1233,"url":"https:\/\/wade.one\/blog\/2026\/04\/24\/ai-coding-agents-are-becoming-normal-developer-infrastructure\/","url_meta":{"origin":1294,"position":0},"title":"AI Coding Agents Are Becoming Normal Developer Infrastructure","author":"","date":"April 24, 2026","format":false,"excerpt":"AI coding agents are moving from novelty demos into normal developer infrastructure. The useful question now is how teams manage them properly.","rel":"","context":"In &quot;AI&quot;","block_context":{"text":"AI","link":"https:\/\/wade.one\/blog\/category\/ai\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":1103,"url":"https:\/\/wade.one\/blog\/2023\/03\/26\/google-bard-vs-chatgpt-a-quick-review\/","url_meta":{"origin":1294,"position":1},"title":"Google Bard vs ChatGPT &#8211; a quick review","author":"Wade","date":"March 26, 2023","format":false,"excerpt":"I am loving the evolution of AI in the last year: it's really developing rapidly and, if you still know how to rephrase your requests to work around limitations of the current implementations you can get a lot out of it. Google Bard and ChatGPT are two of the most\u2026","rel":"","context":"In &quot;AI&quot;","block_context":{"text":"AI","link":"https:\/\/wade.one\/blog\/category\/ai\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":1106,"url":"https:\/\/wade.one\/blog\/2023\/03\/26\/google-bard-php-developer-viewpoint\/","url_meta":{"origin":1294,"position":2},"title":"Google Bard &#8211; PHP developer viewpoint","author":"Wade","date":"March 26, 2023","format":false,"excerpt":"So I asked Bard to give me a reason to help me as as PHP developer then digged in to each point. Here's the answers and when I tried: Bard can help a PHP developer in a number of ways, including (wht: Code generation:\u00a0Bard can generate code in a variety\u2026","rel":"","context":"In &quot;AI&quot;","block_context":{"text":"AI","link":"https:\/\/wade.one\/blog\/category\/ai\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":1155,"url":"https:\/\/wade.one\/blog\/2026\/03\/26\/ai-is-not-a-bubble-denial-is-the-real-problem\/","url_meta":{"origin":1294,"position":3},"title":"AI Is Not a Bubble. Denial Is the Real Problem","author":"Wade","date":"March 26, 2026","format":false,"excerpt":"AI tooling is not going away. What a lot of people are calling a bubble looks more like denial in the face of tools that are already becoming normal developer infrastructure.","rel":"","context":"In &quot;AI&quot;","block_context":{"text":"AI","link":"https:\/\/wade.one\/blog\/category\/ai\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":1158,"url":"https:\/\/wade.one\/blog\/2026\/03\/26\/why-senior-engineers-still-matter-more-in-the-ai-era\/","url_meta":{"origin":1294,"position":4},"title":"Why Senior Engineers Still Matter More in the AI Era","author":"","date":"March 26, 2026","format":false,"excerpt":"AI can generate code faster, but that only makes direction, judgment, and accountability more important. Senior engineers still matter because the expensive mistakes are rarely typing mistakes.","rel":"","context":"In &quot;AI&quot;","block_context":{"text":"AI","link":"https:\/\/wade.one\/blog\/category\/ai\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]},{"id":1182,"url":"https:\/\/wade.one\/blog\/2026\/04\/14\/why-kotlin-coroutines-are-worth-the-learning-curve\/","url_meta":{"origin":1294,"position":5},"title":"Why Kotlin Coroutines Are Worth the Learning Curve","author":"","date":"April 14, 2026","format":false,"excerpt":"Kotlin coroutines are worth learning because they make async code easier to read and easier to reason about. They still need discipline, but they usually beat callback-shaped code by a long way.","rel":"","context":"In &quot;Programming&quot;","block_context":{"text":"Programming","link":"https:\/\/wade.one\/blog\/category\/programming\/"},"img":{"alt_text":"","src":"","width":0,"height":0},"classes":[]}],"_links":{"self":[{"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/posts\/1294","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/comments?post=1294"}],"version-history":[{"count":0,"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/posts\/1294\/revisions"}],"wp:attachment":[{"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/media?parent=1294"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/categories?post=1294"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/wade.one\/blog\/wp-json\/wp\/v2\/tags?post=1294"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}