cd /news/large-language-models/weightbench-evaluating-language-mode… · home › topics › large-language-models › article
[ARTICLE · art-170757] src=arxiv.org ↗ pub= topic=large-language-models verified=true sentiment=· neutral

WeightBench: Evaluating Language Models as Weight-Space Programmers

A November 13, 2026 arXiv paper introduces WEIGHTBENCH, a benchmark that evaluates large language models on 42 Weight-Space Programming (WSP) tasks across five tracks: reading, fixing and writing weights, recognizing impossible edits, and data-free compression, with networks ranging from 2K-parameter control policies to 7B-parameter language models. On all 42 tasks, Claude Opus 5.5 reaches a macro score of 0.535, ahead of Sonnet 5.5 at 0.466, Haiku 5.5 at 0.402 and DeepSeek v4-flash at 0.215; on the 38 tasks it was run on, GPT-6.1-sol reaches 0.568 against Opus's 0.546. The authors report that frontier agents can already read small networks and write working control policies without running them, but cannot yet patch language models, compress them without data, or distinguish an impossible edit from a hard one, and they released the code, task registry, OS-level sandbox harness and live leaderboard publicly on GitHub.

read2 min views4 publishedNov 16, 2026
WeightBench: Evaluating Language Models as Weight-Space Programmers
Image: source
  [Submitted on 13 Nov 2026]


[View PDF](http://arxiv.org/pdf/2611.18553v1)

[HTML (experimental)](https://arxiv.org/html/2611.18553v1)

Abstract:Recent work shows that large language models (LLMs) can operate directly on the parameters of other neural networks. They can read what a network was trained to do, fix a network that fails, and write working networks from scratch, all without gradient descent. This capability has been termed Weight-Space Programming (WSP). Exploring this new paradigm systematically requires a robust benchmark. We introduce WEIGHTBENCH, a benchmark that evaluates LLMs of increasing capability on 42 WSP tasks across five tracks: reading, fixing and writing weights, recognizing edits that are impossible, and data-free compression. The networks whose weights the LLMs read and edit range from 2K-parameter control policies to 7B-parameter language models. Experimental benchmark results show that frontier agents can already read small networks and write working control policies without ever running them, but cannot yet patch language models, compress them without data, or tell an impossible edit from a hard one. On all 42 tasks, Claude Opus 5.5 reaches a macro score of 0.535 (Sonnet 5.5 0.466, Haiku 5.5 0.402, DeepSeek v4-flash 0.215); on the 38 tasks it was run on, GPT-6.1-sol reaches 0.568 against Opus's 0.546. Code, the task registry, our OS-level sandbox harness, and the live leaderboard are publicly available at this https URL.

Additional Features

References & Citations

...

Bibliographic Explorer

(What is the Explorer?) Connected Papers

(What is Connected Papers?) Litmaps

(What is Litmaps?) scite Smart Citations

(What are Smart Citations?) alphaXiv

(What is alphaXiv?) CatalyzeX Code Finder for Papers

(What is CatalyzeX?) DagsHub

(What is DagsHub?) Gotit.pub

(What is GotitPub?) Hugging Face

(What is Huggingface?) ScienceCast

(What is ScienceCast?) Influence Flower

(What are Influence Flowers?) CORE Recommender

(What is CORE?) IArxiv Recommender

(What is IArxiv?) arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

── more in #large-language-models 4 stories · sorted by recency
── more on @weightbench 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/weightbench-evaluati…] indexed:0 read:2min 2026-11-16 · —