@rohanpaul_ai: A new Stanford paper just proved that decentralized multi-agent systems can beat Claude Code and Codex on complex codin…
Summary
A Stanford paper introduces DeLM, a decentralized multi-agent framework that removes the central coordinator in favor of a shared context and task queue, reportedly running up to 2.49× faster and gaining up to +19.2 points in accuracy over Claude Code and Codex baselines on Terminal-Bench 4.0, DeepSWE v1.1, and ProgramBench. Code, 720 trajectories, and an open-source plugin for Codex and Claude Code are released.
View Cached Full Text
Cached at: 10/08/26, 04:19 PM
A new Stanford paper just proved that decentralized multi-agent systems can beat Claude Code and Codex on complex coding tasks, running up to 2.49× faster!
DeLM tackles a key bottleneck in multi-agent systems: time wasted repeating work and waiting on other agents.
DeLM replaces the central coordinating agent with a shared context and task queue. Agents pick up tasks independently, share discoveries as they happen, and build on each other’s progress. When one agent finds a solution or hits a dead end, the others can use that information immediately.
DeLM beats state-of-the-art in both speed and accuracy on long-horizon tasks from Terminal-Bench 4.0, DeepSWE v1.1, and ProgramBench:
- Up to 2.49× faster execution than the vanilla Claude Code and Codex baselines
- Up to +19.2 percentage points in accuracy over the vanilla Claude Code and Codex baselines
- Up to +19.9 points in ProgramBench test pass rate within the same 120-minute budget
Built on Codex and Claude Code. Code and 720 trajectories are available so you can explore how the agents collaborate! An open-source plugin lets you try DeLM directly in Codex and Claude Code.
Similar Articles
Decentralized Multi-Agent Systems with Shared Context
This paper introduces Decentralized Language Models (DeLM), a framework for multi-agent systems that uses parallel agents with a shared verified context to improve test-time scaling and reduce costs, achieving state-of-the-art results on SWE-bench Verified and LongBench-v2.
Open-sourced a director/implementer delegation protocol for coding agents (Claude Code + Codex)
Open-sourced a platform-neutral director/implementer delegation protocol for coding agents like Claude Code and Codex, with schemas, adapters, and example scenarios to address trust in subagent completion reports.
I built an open-source multi-agent SDLC harness that beats a cold Claude Code run on large repos, by learning the repo once. Real benchmarks (incl. where it loses) inside. [P]
AutoDev Studio is an open-source multi-agent AI coding harness that reduces costs by building a persistent repository knowledge base, outperforming cold-start Claude Code runs on large repos in most tasks.
@yoheinakajima: more ppl are now trying out this approach of agents communicating with a shared state (vs talking to each other)
Azalia Mirhoseini highlights DeLM, a decentralized language model approach where agents communicate via shared state, achieving ~10% improvement on SWE-bench Verified with Gemini-3 Flash at less than half the cost.
Compiler backed agent beats Claude Code, OpenCode & other major agents and harnesses (benchmarks linked)
Benzi is a compiler-backed AI coding agent that uses deterministic static analysis and code indexing instead of reading raw source code, reporting 78.2% on SWE-bench Verified for under 10 cents per fix while reading far fewer tokens than Claude Code, DeepSeek's harness, and OpenCode.