OpenAI officially released GPT-5.5 on April 23, 2026, positioning the model as a significant advancement in agentic computing. The company claims the model is designed to navigate software environments, manage multi-part tasks, and perform complex operations such as coding, data analysis, and online research with minimal human intervention.
According to OpenAI, GPT-5.5 maintains the same per-token latency as its predecessor, GPT-5.4, while demonstrating improved reasoning and efficiency. The company reports that the model uses fewer tokens to complete equivalent Codex tasks, allowing for higher performance without sacrificing speed in real-world serving environments.
The model is currently available to Plus, Pro, Business, and Enterprise users within ChatGPT and Codex. OpenAI noted that API deployments for GPT-5.5 are subject to distinct safety and security requirements, and the company is currently working with partners to finalize those protocols for broader scaling.
To address potential risks, OpenAI subjected the model to its full suite of safety and preparedness frameworks before release. This process included internal and external red-teaming, as well as feedback from nearly 200 trusted early-access partners who tested the model against real-world use cases.
In conjunction with the release, OpenAI launched a specialized Bio Bug Bounty program focusing on GPT-5.5 within the Codex Desktop environment. The program invites security and biosecurity experts to attempt to find a “universal jailbreak” that can successfully answer five specific bio-safety questions without triggering the model’s moderation filters.
OpenAI is offering a $25,000 reward to the first researcher who successfully clears all five questions, with the possibility of smaller awards for partial findings. The program will accept applications through June 22, 2026, with active testing scheduled to occur between April 28 and July 27, 2026.
Participants in the bounty program are required to sign a non-disclosure agreement (NDA) regarding their findings and communications. The company stated that the bounty is part of its broader effort to bolster safeguards against the misuse of advanced AI capabilities in biology and cybersecurity.
OpenAI’s system card for the release indicates that GPT-5.5 Pro, which utilizes parallel test-time compute, is being evaluated separately from the base model. The company determined that the Pro setting could potentially alter the model’s risk profile, necessitating a distinct safety and safeguards posture.
