Qwen3.8-27B

Mon, 17 Aug 2026 00:00:00 +0000

Alibaba’s Qwen team released Qwen3.8-27B on August 14, 2026, with open weights available immediately under the Apache 2.0 license. It’s a dense 27 billion parameter model rather than a mixture-of-experts design, and it handles text, images, and video, including diagrams, documents, and multi-hour footage. Context runs to 262,000 tokens natively and can stretch to a million using the YaRN scaling method. Alibaba says it beats the larger Qwen3.7-Plus on coding and office tasks and has sharpened its agentic planning, letting it complete multi-step tasks with less supervision.

Poolside Laguna XS 2.1

Sun, 09 Aug 2026 00:00:00 +0000

Laguna XS 2.1 is Poolside’s smallest open-weight coding model, launched July 2, 2026 as a free download on Hugging Face and OpenRouter ahead of the larger Laguna S 2.1 release later that month. It’s a 33 billion parameter mixture-of-experts model small enough to run on a single GPU while still handling multi-step software engineering tasks, and it ships with DFlash speculator models to speed up inference. Like the rest of the Laguna line it’s released under OpenMDW-1.1, a permissive license Poolside designed specifically for AI model weights, aimed at developers who want an agentic coding model they can self-host without a large hardware budget.

Single-Gpu on Best of AI

Qwen3.8-27B

Poolside Laguna XS 2.1