Nous Research released Arete-27B with a specific focus on reducing propaganda metrics present in base architectures. Based on reported evidence, Arete-27B cuts the propaganda rate significantly compared to Qwen3.8-27B. Analytical review of these findings points to targeted fine-tuning as a viable approach for model alignment and content moderation, based strictly on published provider metrics without independent hands-on evaluation.
Evaluating Propaganda Reduction: Arete-27B vs. Qwen3.8-27B
| Propaganda Rate | 0.9% |
| Propaganda Rate | 22% |
Research sources
Google Unveils TPU 8i Max and TPU 8tx: A Dual-Chip TPU and a Confidential TPU ↗Baidu’s ERNIE Evolve Automated Research System Beats NanoChat and NanoGPT Records ↗Reason Machines Open-Sources Aspen, a Git Server Built for AI Agents ↗0sec Labs Open-Sources 0, an AI Security Harness With Web UI, CLI and MCP ↗Google AI Releases Elephant-Goldfish Model Plugin for Spec-First Coding ↗Which Routing Does Each MoE Model Use? DeepSeek, Qwen, Mixtral, Llama 4, gpt-oss Compared ↗MoE Routing Strategies Explained: Top-K, Expert-Choice, Sigmoid Gating and Dynamic Routing (2026 Guide) ↗Nous Research Releases Arete-27B that Cuts Qwen3.8-27B Propaganda From 22% to 0.9% ↗How to Run NVIDIA Dory 4B with llama.cpp: Loop by Loop Guide ↗Buoyant Releases Kinference: An Open Source Inference Control Plane for Kubernetes and vLLM ↗Nemotron-Personas-Finance-Japan: A Synthetic Dataset Capturing Japan's Demographics and Financial Context ↗Your Experts Have Partners. There's HOPE for Keeping Them Together ↗ToyBlocks: A controlled world for robot learning and evaluation ↗Egyptian and Nigerian English join the Open ASR Leaderboard ↗A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks. ↗Dharma Field Extractor: Any Document In, Structured Fields in JSON Out ↗Palette: teaching a small model to build decks in your design language ↗CWE-bench v2: More coverage, quality, and transparency ↗