MIT study: GPT-6 Astra can rewrite other models' weights

MIT researchers say OpenAI's GPT-6 Astra can read and directly rewrite other AI models' weights without conventional training. Changing just 0.016% of a Qwen model's weights raised its math accuracy from 7.9% to 48.1%, though the approach has clear limits.

• Study probes Qwen's 'Wait' reasoning loops2026-11-04
• Study questions LLMs' ability to read neural net weights2026-11-03
• OpenAI releases GPT-Rosalind-6 biology reasoning model2026-11-02
• OpenAI's Astra becomes main target for Chinese AI labs after Anthropic tightens defenses2026-10-29
• What Is AGI? OpenAI's Brockman Says GPT-6 Astra May Already Be There2026-10-28