Yhn HealthMedicine first seen 1 d ago, last 1 h ago, peak #3

Memdoor team of five Qwen models beats Fable 5 on coding benchmark

Original: Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hard

An open-source project called Memdoor reports that a team of five Qwen3.8-27B language models working together outperformed Fable 5 on LiveCodeBench Hard, a demanding programming benchmark. The result, published on GitHub, suggests that coordinating multiple mid-size models can rival stronger single models on difficult coding tasks, drawing attention from developers interested in open-model performance.

Why now: Open-source multi-model setups beating a top closed model on a hard benchmark is a notable and surprising result for the AI community

MemdoorQwen3.8-27BFable 5LiveCodeBenchGitHub

Open on hn →

Rank over time, top of the chart is #1. 117 snapshots from 1 d ago to 1 h ago.

Evidence

• Memdoor: A team of 5 Qwen3.8-27B models beats Fable 5 on LiveCodeBench Hard · guregodevo · 450

API: https://socialmediatrends-api.osmike.com/v1/trends/4425810