OpenAI's global deployment of GPT-5.5 is officially underway. The announcement, made on April 23, 2026, marks the beginning of a phased rollout that will bring OpenAI's most capable model to users worldwide. GPT-5.5 — internally codenamed 'Spud' during development — represents a significant step beyond GPT-5 in agentic capabilities, with OpenAI describing it as 'a new class of intelligence' rather than an incremental update.
The model is available immediately to ChatGPT Pro subscribers at the $200/month tier, with rollout to the new $100/month Pro tier and Plus ($20/month) subscribers to follow. The introduction of the $100/month tier signals OpenAI's intent to capture heavy users who find Plus insufficient but balk at the $200 price point.
What Makes GPT-5.5 Different
GPT-5.5 is designed as an agentic model: one that can plan and execute multi-step tasks, use tools, browse the web, write and run code, and coordinate with other AI systems over extended periods. OpenAI describes it as capable of operating autonomously for hours on complex tasks, with minimal human intervention.
The model's pricing reflects its positioning as a premium product: $5/M input tokens and $30/M output tokens via the API. This makes it the most expensive model in OpenAI's lineup, but also the most capable for agentic workloads. For comparison, GPT-5.4 costs $2.50/M input and $15/M output — the 2x price premium for GPT-5.5 is justified, according to OpenAI, by substantially improved performance on agentic benchmarks.
Security researchers have already begun probing GPT-5.5's capabilities. Early reports suggest the model shows significant improvements in vulnerability detection compared to its predecessors, outperforming GPT-5.4 in both black-box and white-box security testing scenarios. The competitive context is intense: Anthropic's Claude Opus 4.7 is widely regarded as GPT-5.5's closest competitor, and DeepSeek V4-Pro — released the same day — offers competitive performance at a fraction of the price.