{"id":75715,"date":"2026-04-24T09:03:51","date_gmt":"2026-04-24T02:03:51","guid":{"rendered":"https:\/\/hbbgroup.net\/openai-releases-gpt-5-5-faster-smarter-and-pricier\/"},"modified":"2026-04-24T09:03:51","modified_gmt":"2026-04-24T02:03:51","slug":"openai-releases-gpt-5-5-faster-smarter-and-pricier","status":"publish","type":"post","link":"https:\/\/hbbgroup.net\/zh\/openai-releases-gpt-5-5-faster-smarter-and-pricier\/","title":{"rendered":"OpenAI Releases GPT-5.5: Faster, Smarter\u2014And Pricier"},"content":{"rendered":"<div>\n<div>\n<h4 color=\"#333\">In brief<\/h4>\n<ul>\n<li>GPT-5.5 launches today for Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, with API access coming soon at $5\/M input tokens and $30\/M output tokens.<\/li>\n<li>The model achieves 82.7% on Terminal-Bench 2.0\u2014a benchmark testing complex command-line workflows\u2014beating Claude Opus 4.7 at 69.4% and Gemini 3.1 Pro at 68.5%.<\/li>\n<li>GPT-5.5 uses significantly fewer tokens to complete the same Codex tasks as GPT-5.4, making it more efficient despite being priced higher.<\/li>\n<\/ul>\n<\/div>\n<p>OpenAI launched GPT-5.5 on Thursday, pitching it essentially as a model targeted at agentic computer use. It writes and debugs code, browses the web, fills out spreadsheets, and keeps working through multi-step tasks without needing a human to babysit every move.<\/p>\n<p>The release is already rolling out today to Plus, Pro, Business, and Enterprise subscribers across ChatGPT and Codex, OpenAI said.<\/p>\n<p>\u201cWe\u2019re releasing GPT\u20115.5, our smartest and most intuitive-to-use model yet, and the next step toward a new way of getting work done on a computer,\u201d OpenAI <a href=\"https:\/\/openai.com\/index\/introducing-gpt-5-5\/\" target=\"_blank\" rel=\"nofollow external noopener\">said<\/a> in an announcement. \u201cThe gains are especially strong in agentic coding, computer use, knowledge work, and early scientific research\u2014areas where progress depends on reasoning across context and taking action over time.<\/p>\n<div>\n<blockquote>\n<p lang=\"en\" dir=\"ltr\">Introducing GPT-5.5<\/p>\n<p>A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done.<\/p>\n<p>Now available in ChatGPT and Codex. <a href=\"https:\/\/t.co\/rPLTk99ZH5\" data-wpel-link=\"internal\">pic.twitter.com\/rPLTk99ZH5<\/a><\/p>\n<p>\u2014 OpenAI (@OpenAI) <a href=\"https:\/\/twitter.com\/OpenAI\/status\/2047376561205325845?ref_src=twsrc%5Etfw\" data-wpel-link=\"internal\">April 23, 2026<\/a><\/p>\n<\/blockquote>\n<\/div>\n<p>The big headline from OpenAI: GPT-5.5 is measurably smarter than its predecessor, GPT-5.4\u2014and it&#8217;s not slower. Matching GPT-5.4&#8217;s per-token latency in real-world serving while hitting higher scores across benchmarks is the kind of efficiency improvement that usually doesn&#8217;t happen. Bigger models tend to be slower when running under the same hardware.<\/p>\n<p>On Terminal-Bench 2.0, which tests how well a model handles complex command-line workflows that require planning and iterative tool use, GPT-5.5 scores 82.7%. Claude Opus 4.7 lands at 69.4%, while Gemini 3.1 Pro sits at 68.5%. That&#8217;s not a marginal lead.<\/p>\n<p>On GDPval, a benchmark testing knowledge work across 44 real occupations\u2014from finance to legal research to product management\u2014GPT-5.5 matches or beats industry professionals in 84.9% of comparisons.<\/p>\n<figure><img loading=\"lazy\" alt=\"GPT 5.5 benchmarks, compared against other models from OpenAI, Gemini and Claude from Anthropic\" width=\"2380\" height=\"1118\" decoding=\"async\" data-nimg=\"1\" src=\"https:\/\/img.decrypt.co\/insecure\/rs:fit:3840:0:0:0\/plain\/https:\/\/cdn.decrypt.co\/wp-content\/uploads\/2026\/04\/Captura-de-pantalla-2026-04-23-a-las-15.36.45.png@webp\"><figcaption>Image: OpenAI<\/figcaption><\/figure>\n<p>It\u2019s also a pretty good coder, as expected. On Expert-SWE, an internal benchmark for long-horizon coding tasks with a median estimated human completion time of 20 hours, GPT-5.5 outperforms GPT-5.4. On SWE-Bench Pro, which grades real-world GitHub issue resolution, it reaches 58.6%. Claude Opus 4.7 scores higher at 64.3%, but OpenAI claims it may be because \u201cAnthropic reported\u2060 signs of memorization on a subset of problems\u201d<\/p>\n<p>This launch lands in a market that is moving rapidly since the boom of agentic AI. <a href=\"https:\/\/decrypt.co\/360148\/openai-launches-gpt-5-4-quitgpt-exodus-gains-steam\" target=\"_blank\">GPT-5.4 arrived just two days after GPT-5.3<\/a>, while Xiaomi went from <a href=\"https:\/\/decrypt.co\/362633\/xiaomi-mimo-v2-pro-review-so-good-mistaken-deepseek-v4\" target=\"_blank\">MiMo-V2-Pro<\/a> to <a href=\"https:\/\/decrypt.co\/365184\/xiaomi-mimo-2-5-pro-ai-see-hear-act-one-model\" target=\"_blank\">MiMo 2.5 Pro<\/a>\u2014with full multimodal capabilities\u2014in roughly five weeks. The gap between GPT-5.4 and GPT-5.5 was about seven weeks. That&#8217;s the tempo now.<\/p>\n<p>But will this model make a difference for everyday users who are not always coding the next big thing? If you&#8217;re on a free tier, no: GPT-5.5 isn&#8217;t coming to free users. If you&#8217;re paying for Plus at $20\/month, it rolls out today. We tried testing it under our Pro account, but the model was not immediately available.<\/p>\n<div>\n<figure><img loading=\"lazy\" alt width=\"2466\" height=\"1704\" decoding=\"async\" data-nimg=\"1\" src=\"https:\/\/img.decrypt.co\/insecure\/rs:fit:3840:0:0:0\/plain\/https:\/\/cdn.decrypt.co\/wp-content\/uploads\/2026\/04\/Captura-de-pantalla-2026-04-23-a-las-15.50.57.png@webp\"><\/figure>\n<\/div>\n<p>The bigger deal is probably what GPT-5.5 does inside Codex\u2014OpenAI&#8217;s agentic coding environment\u2014where it is proven to be more powerful. \u201cIt genuinely feels like I\u2019m working with a higher intelligence, and there\u2019s almost a sense of respect,\u201d Pietro Schirano, CEO of MagicPath, said in a quote shared by OpenAI.<\/p>\n<p>GPT-5.5 Pro, designed for harder, higher-accuracy work, is rolling out separately to Pro, Business, and Enterprise users in ChatGPT. On BrowseComp, which tests a model&#8217;s ability to track down hard-to-find information across the web, GPT-5.5 Pro scores 90.1%, ahead of Gemini 3.1 Pro at 85.9%.<\/p>\n<p>The model is also the most intelligent on average based on the Artificial Analysis Index. GPT 5.5 reports a more efficient and useful use of tokens, producing better results in general.<\/p>\n<figure><img loading=\"lazy\" alt=\"Artificial Analysis Index\" width=\"1598\" height=\"1090\" decoding=\"async\" data-nimg=\"1\" src=\"https:\/\/img.decrypt.co\/insecure\/rs:fit:3840:0:0:0\/plain\/https:\/\/cdn.decrypt.co\/wp-content\/uploads\/2026\/04\/Artificial-Analysis-Intelligence-Index.png@webp\"><figcaption>Image: OpenAI<\/figcaption><\/figure>\n<p>The pricing, however, could shock some users. The API will charge $5 per million input tokens and $30 per million output tokens when it launches, which OpenAI says is coming &#8220;very soon.&#8221; GPT-5.5 Pro in the API will cost $30 per million input tokens and $180 per million output tokens.<\/p>\n<p>Those figures are higher than GPT-5.4\u2014$2.50 per million tokens of input and $15.00 per million tokens of output\u2014while pricing for GPT-5.5 Pro remains the same as GPT-5.4 Pro.<\/p>\n<p>That said, OpenAI CEO Sam Altman <a href=\"https:\/\/x.com\/sama\/status\/2047379036419014928\" target=\"_blank\" rel=\"nofollow external noopener\">argued on X<\/a> that token efficiency gains offset the cost\u2014GPT-5.5 completes the same Codex tasks with fewer tokens, which means cheaper runs even at a higher per-token rate.<\/p>\n<p>Just for comparison, Xiaomi MiMo v2.5 Pro charges $1 and $3 per million tokens of input and output, Minimax M2.7 costs $0.30 and $1.20 respectively, and Kimi K2.5 requires $0.44 and $2.00 per million tokens.<\/p>\n<div>\n<h3>Daily Debrief Newsletter<\/h3>\n<p>Start every day with the top news stories right now, plus original features, a podcast, videos and more.<\/p>\n<\/div>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>In brief GPT-5.5 launches today for Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, with API access coming [&hellip;]<\/p>","protected":false},"author":5,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[],"class_list":["post-75715","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"acf":[],"_links":{"self":[{"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/posts\/75715","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/comments?post=75715"}],"version-history":[{"count":0,"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/posts\/75715\/revisions"}],"wp:attachment":[{"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/media?parent=75715"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/categories?post=75715"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hbbgroup.net\/zh\/wp-json\/wp\/v2\/tags?post=75715"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}