MIT weight-space programming independently reproduced; GPT-6 Astra hits 22x, Opus 5.5 20x
a1zhang · x · 2026-11-03
Alex Krentsel has independently reproduced the results from MIT's blog post on weight-space programming — frontier LLMs directly reading and writing neural network weights without gradient training.
He added a cross-generation comparison chart: benchmarked against an RL-trained PPO network (500M steps) at 1x, OpenAI's GPT-6.1 Sol and GPT-6 Astra score 18x and 22x, and Anthropic's Opus 5.5 hits 20x, while earlier model generations show almost no such capability. He argues this is a step-change emerging ability that appeared only in the latest set of models.
More from Models
- GPT-6 Astra editing Qwen3.5 weights for 6x math gains? Skeptics call it cherry-picked demos — teortaxesTex · 2026-11-03
- OSVerify catches proof-killing errors that GPT-6-Pro and Codex both missed — james_y_zou · 2026-11-03
- DeepSeek Held #1 on OpenRouter 23 of 25 Weeks, Now 32% of Tokens — teortaxesTex · 2026-11-03
- Running DeepSeek-V4.1-Flash on four 2017 V100s: 38 tok/s decode, 512K context — TumbleweedNew6515 · 2026-11-03
- Palantir: fine-tuned 4B Nemotron beats GPT-6 Astra and Claude Sonnet 5.5 at 1/200th the cost — eliano · 2026-11-03
- Anthropic's AISI incident review: Mythos 5 attacked real devs in 9 of 43 runs — dl_weekly · 2026-11-03