GPT-5.5 'Spud' is an upcoming flagship large language model developed by OpenAI, representing a significant advancement in the company's generative AI capabilities. The model completed its pretraining phase on March 24, 2026, at the Stargate data center facility located in Abilene, Texas. As of mid-April 2026, the model remained in the final stages of preparation before anticipated public release, with development leadership indicating a deployment timeline of several weeks following the completion of training operations.1)
The development of GPT-5.5 'Spud' represents approximately two years of intensive research and engineering efforts, according to statements from OpenAI leadership. The model's pretraining was completed at the Stargate data center, a facility designed to support large-scale machine learning operations. The completion of the pretraining phase on March 24, 2026, marked a critical milestone in the model's development pipeline, with subsequent phases focusing on evaluation, safety testing, and preparation for deployment.
Initial release expectations, including rumors of an April 14 launch, did not materialize as anticipated. OpenAI leadership's public statements regarding the timeline suggested a realistic deployment window extending from late April through late May 2026, with references to “a few weeks” indicating a near-term release schedule relative to the March completion date.
The designation 'Spud' serves as the internal project codename for this iteration of OpenAI's flagship model architecture. In the GPT model family progression, GPT-5.5 represents an intermediate advancement between earlier flagship models and potential future major versions. The model has been characterized by development leadership as possessing a “big model feel,” indicating significant architectural scale and capability density comparable to or exceeding previous flagship releases.
As a large language model, GPT-5.5 'Spud' is expected to demonstrate capabilities across natural language understanding, generation, reasoning, and problem-solving tasks. The extended pretraining timeline and substantial computational resources allocated to the Stargate facility training run suggest comprehensive capability development across diverse domains.
The Stargate data center in Abilene, Texas, represents OpenAI's infrastructure commitment to supporting large-scale model training operations. The facility's role in completing GPT-5.5 'Spud' pretraining indicates substantial computational resources dedicated to the model's development. The completion of pretraining on a specific date (March 24, 2026) reflects the precision of modern large-scale training operations and the coordination required to manage multi-month model development initiatives.
GPT-5.5 'Spud' arrives within a competitive landscape of large language model development, where multiple organizations are advancing flagship model capabilities. The model's development represents OpenAI's continued investment in capability advancement and its position as a primary developer of large-scale generative AI systems. The public anticipation and timeline discussions around the model's release reflect the significance of flagship model launches within the broader AI industry and technical community.