OpenAI unveils GPT-5.5 for ChatGPT and Codex, pricier

OpenAI released GPT-5.5 on April 23, rolling the model out to Plus, Pro, Business and Enterprise subscribers in ChatGPT and to Codex users. Free-tier accounts will not receive the model. OpenAI said API access will arrive soon with new per-token rates.

OpenAI describes GPT-5.5 as aimed at desktop and coding workflows that require multi-step action and tool use. The company reports the model can write and debug code, browse the web, fill spreadsheets and carry tasks forward with less human supervision. A higher-accuracy Pro variant is available for Pro, Business and Enterprise customers in ChatGPT.

OpenAI released benchmark results showing gains over prior versions and some competitors. On Terminal-Bench 2.0, which tests complex command-line workflows and iterative tool use, GPT-5.5 scored 82.7%, compared with 69.4% for Claude Opus 4.7 and 68.5% for Gemini 3.1 Pro. On GDPval, a test covering 44 occupations such as finance and legal research, GPT-5.5 matched or beat industry professionals in 84.9% of comparisons. The Pro model scored 90.1% on BrowseComp, a browsing and research test, versus 85.9% for Gemini 3.1 Pro.

In coding tests, OpenAI reported improvements over GPT-5.4 on long-horizon engineering tasks measured by Expert-SWE and stronger results on some GitHub issue resolution measures. On SWE-Bench Pro, GPT-5.5 reached 58.6%, while Anthropic’s Claude Opus 4.7 scored 64.3%. OpenAI noted Anthropic had reported signs of memorization on a subset of problems.

OpenAI said GPT-5.5 matches GPT-5.4’s per-token latency in real-world serving and completes equivalent Codex tasks using fewer tokens. The company argues that lower token usage can offset higher per-token charges. OpenAI CEO Sam Altman wrote on the company feed that token-efficiency improvements can reduce total run costs despite higher per-token prices.

API pricing that OpenAI plans to introduce is $5 per million input tokens and $30 per million output tokens. A Pro API tier will carry rates of $30 per million input tokens and $180 per million output tokens. For reference, GPT-5.4’s public API pricing was $2.50 per million input tokens and $15 per million output tokens.

Pietro Schirano, chief executive of MagicPath, described the user experience as feeling like working with a higher intelligence and noted an unexpected sense of respect in interactions, a quote OpenAI shared.

OpenAI said the release prioritizes paid subscriptions and enterprise customers. The company reported that the gap between GPT-5.4 and GPT-5.5 was about seven weeks and that some users did not see immediate access during early rollout windows for Pro accounts. OpenAI emphasized that token efficiency will be a key factor for customers evaluating the model against per-token pricing.