---
format: "aidr-story-markdown/v1"
id: "2b8b90df05129da48cdd091b71a71856c1a895ab98451147588fb4e9dd497e82"
canonical_url: "https://aidr.today/2b8b90df?lang=en"
title: "Stanford DeLM Beats Claude Code Accuracy by 17.5 Percentage Points"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-10-08T20:54:45.000Z"
category: "Frameworks"
topics: ["agent","framework"]
source_urls: []
summary: "Stanford researchers released an updated version of DeLM, a decentralized framework for AI agents that outperforms Codex and Claude Code on several benchmarks. Utilizing GPT-6-Astra and Claude Opus 5.5, the system improved accuracy by up to 17.5 percentage points over the strongest baseline."
---

# Stanford DeLM Beats Claude Code Accuracy by 17\.5 Percentage Points

> [Open the canonical story](<https://aidr.today/2b8b90df?lang=en>)

**Published:** 2026-10-08T20:54:45.000Z
**Category:** Frameworks
**Topics:** agent, framework

## Summary

Stanford researchers released an updated version of DeLM, a decentralized framework for AI agents that outperforms Codex and Claude Code on several benchmarks\. Utilizing GPT\-6\-Astra and Claude Opus 5\.5, the system improved accuracy by up to 17\.5 percentage points over the strongest baseline\.

## Sources

_No safe source URL was recorded._