OpenAI officially launched GPT-5.5 on April 24, 2026, introducing what the company describes as its most advanced artificial intelligence model to date. The release marks a significant shift from conversational interfaces toward agentic systems capable of autonomous execution across software environments. According to official statements, GPT-5.5 is engineered to handle complex, multi-step tasks in coding, scientific research, and general computer use with minimal human intervention.
The new model demonstrates substantial improvements in technical benchmarks. OpenAI reported that GPT-5.5 achieved an 82.7% accuracy rate on Terminal-Bench 2.0, a metric for console-based tasks and tool coordination, surpassing the 75.1% recorded by its predecessor, GPT-5.4. In software engineering, the model scored 73.1% on the Expert-SWE benchmark, which evaluates performance on long-horizon coding tasks. Furthermore, the company noted that GPT-5.5 can resolve real-world GitHub issues in 58.6% of cases, representing a major leap in autonomous development capabilities.
OpenAI President Greg Brockman highlighted the model's ability to navigate ambiguity, stating that GPT-5.5 can analyze unclear problems and determine necessary subsequent steps without explicit subtask instructions. CEO Sam Altman characterized the release as a new class of intelligence designed for real work, emphasizing its capacity to plan, use tools, and verify its own output. The model also introduces a 20% increase in token generation speed and improved token efficiency, allowing it to complete tasks with lower computational overhead than previous versions.
GPT-5.5 is available immediately to subscribers of ChatGPT Plus, Pro, Business, and Enterprise plans. A specialized high-precision version, GPT-5.5 Pro, is reserved for Pro, Business, and Enterprise tiers to support demanding workloads in legal, scientific, and financial sectors. For developers, the model features a context window of up to one million tokens via the OpenAI API. Standard API pricing is set at $5 per million input tokens and $30 per million output tokens, while the Pro tier carries a higher rate of $30 for input and $180 for output.
The launch includes strategic integrations with industry partners. Snowflake announced same-day availability of GPT-5.5 on its Cortex AI platform, enabling enterprise users to build autonomous agents within secure data perimeters. The release follows a period of heavy infrastructure investment by OpenAI and Microsoft, which has expanded its capital expenditures to support the compute requirements of the new model architecture. While the model retakes the lead in several public benchmarks, it enters a competitive market alongside recent releases from Anthropic and Google.