AI Self-Improvement
The ability of a model to participate in improving itself or models of the same kind, for example by assisting researchers in writing training code, designing evaluation schemes, or optimizing data pipelines; it is regarded as one of the key paths toward more capable AI.
Timeline (last 90 days)
The OpenAI system card states that GPT-6.2 Astra is the first externally released model to meet the high threshold for AI self-improvement; this statement is OpenAI's own account and has not been verified by a third party.
For the first time, OpenAI rates the AI self-improvement capability of a publicly released model as HIGH.
Recursive views Flint as an early test of AI self-improvement.
All Facts (3)
The OpenAI system card states that GPT-6.2 Astra is the first externally released model to meet the high threshold for AI self-improvement; this statement is OpenAI's own account and has not been verified by a third party.
50%UnverifiedFor the first time, OpenAI rates the AI self-improvement capability of a publicly released model as HIGH.
50%UnverifiedRecursive views Flint as an early test of AI self-improvement.
50%