Unverified60% confidenceFactExact time
In a study titled "Sparks of Weightspace Programming," the MIT team had OpenAI's GPT-6 Astra directly read and modify another model's weights without training or gradient descent.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
11/4/2026
First Seen
Valid until: 2/2/2027
Sources
Related Entities
Related Claims
UnverifiedAccording to OpenAI's Noam Brown, the model involved is similar in scale to GPT-5.6-level, is a model used solely for internal research and trained to be highly persistent and diligent, and is not the next-generation Astra series.75% similarUnverifiedFour MIT researchers (led by Rikia Takehi and Jaden Teo) published a paper proposing the concept of 'weight-space programming,' i.e., handing the raw weights of a small network to GPT-6 Astra and letting it read, repair, or write new weights from scratch.73% similarUnverifiedGPT-6 Astra can determine what task a network was trained to perform simply by looking at its raw weights.71% similarVerifiedUnsloth's OpenAI-compatible API has enabled gpt-6-astra (GPT-6 Astra) by default for Codex login, supporting inference intensity control ranging from Low to Max.71% similarVerifiedGPT-6 Astra is an entirely new pre-trained model generation, not a minor version iteration.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1135393API
curl https://kongchang.com/api/v1/knowledge/claims/1135393MCP
get_claim(id=1135393)