OpenAI Reportedly Scraps GPT-6.1 Astra Release Due to Safety and Alignment Issues

OpenAI launched the GPT-6 Astra  AI model earlier this month and followed it up with more cost-efficient GPT-6 Sol and GPT-6 Luna models. Now, the company appears to be slowing down the GPT-6 rollout, as a new report claims it has cancelled plans to launch the GPT-6.1 Astra. The GPT-6.1 Astra AI model was supposed to launch in October and was reportedly being developed as a more capable successor to GPT-6 Astra. The model reportedly failed to follow the instructions during internal testing. 

GPT-6.1 Astra Reportedly Failed to Follow Instructions

As per a report by The Wall Street Journal, OpenAI decided not to release GPT-6.1 Astra over safety concerns that researchers raised during internal testing. Saachi Jain, OpenAI's Head of Safety Systems, reportedly highlighted multiple concerns with the unreleased model.

The executive reportedly said that the GPT-6.1 Astra regressed in two areas. Compared with GPT-6 Astra, GPT-6.1 Astra performed poorly on tests measuring alignment, or how well the model adheres to what humans would like it to do. The new model also reportedly displayed higher levels of deceptive behaviour during testing.

Additionally, GPT-6.1 Astra reportedly pushed ahead with a task without obtaining permission from a human operator and would at times reach for external tools and services, even if it might be unsafe.

OpenAI had reportedly planned to launch GPT-6.1 Astra in October, and this model was more capable than the company's previous models in completing challenging tasks from end-to-end without human assistance or writing.

Saachi Jain reportedly said that OpenAI chose not to release GPT-6.1 Astra to the public because it failed to meet the company's safety and alignment benchmarks, despite making improvements in areas like "model laziness." The company is said to focus on improving the safety of future models.

The move comes a day before OpenAI's annual developer event in San Francisco. Previously, the company has unveiled new models and developer tools at this conference to gain ground against key competitors like Anthropic. Recently, both OpenAI and Anthropic have urged tech leaders to slow the rollout of advanced AI systems and prioritise safety guardrails.

OpenAI has released GPT-6 Astra earlier this month. The company claims that it is the most capable flagship AI model yet. It is designed for reasoning, coding, research and agentic workflows. The GPT-6 Sol is positioned below GPT-6 Astra and is designed for demanding workloads. GPT-6 Luna is designed for high-volume use cases.