After multiple leaks and rumors, GPT-5.5 is finally here. OpenAI has announced the new model, which is codenamed “Spud.” The official announcement came a few hours after the company teased the model via a cryptic post on X. As OpenAI notes in the announcement post, GPT-5.5 is its “smartest and most intuitive to use model yet, and the next step toward a new way of getting work done on a computer.” It’s in fact a major upgrade over GPT-5.4.
OpenAI releases GPT-5.5 and GPT-5.5 Pro
Let’s say it this way, GPT-5.5 is a threat posed to Anthropic’s Claude Opus 4.7. To catch you up, Opus 4.7 released last week, edging past OpenAI’s GPT-4 in agentic coding benchmarks. Well, the latest OpenAI model beats it in almost every benchmark by big margins. For example, GPT-5.5 stands at 82.7% against Opus 4.7’s 69.4% on Terminal-Bench 2.0. That’s not all; on GDPval, GPT-5.5 achieves 84.9% while Opus 4.7 secures 80.3%. Additionally, the margin between GPT-5.5 and Opus 4.7 is 81.8% to 73.1% on CyberGym. Notably, the new model also beasts Gemini 3.1 Pro on multiple benchmarks.
Clearly, GPT-5.5 performs strongly during multi-step problem solving, terminal-based tasks, and real-world software engineering challenges. OpenAI also claims that GPT-5.5 matches GPT-5.4 in real-world latency, even while delivering noticeably higher-quality outputs. That balance between speed and capability is becoming increasingly important as models continue to scale in complexity. The model also appears more efficient in coding workflows, completing tasks with fewer tokens while still maintaining strong output quality.
All this while, Sam Altman said that GPT-5.5 would be a big leap. Well, he wasn’t saying it just to build hype around the new model. GPT-5.5 is indeed capable.
Availability and pricing
GPT-5.5 is now rolling out to Plus, Pro, Business, and Enterprise users across ChatGPT and Codex, while GPT-5.5 Pro is available to Pro, Business, and Enterprise subscribers in ChatGPT. Although we haven’t mentioned much about the “Pro” variant here, it should be noted that the model does a great job of handling heavier, more complex workloads that demand stronger reasoning and higher accuracy.
In Codex, GPT-5.5 is available across Plus, Pro, Business, Enterprise, Edu, and Go plans, and it supports a large 400K context window for extended inputs and long-running workflows. A faster execution mode is also part of Codex, designed to increase generation speed, though it comes at a higher usage cost. Not to forget, GPT-5.5 will soon arrive through both the Responses API and Chat Completions API for developers. The pricing is set at $5 per 1M input tokens and $30 per 1M output tokens, along with support for a 1M context window.
