---
type: Guide
title: "Best Open-Weight LLMs 2026: Self-Hosting Guide"
description: "Compare the best open-weight LLMs for self-hosting in 2026: Qwen3.5, GLM-5.2, gpt-oss-120b, DeepSeek-V4, Kimi K2.7 Code, Gemma 4, Mistral Small 4 and MiniMax M3 — verified parameter counts, context lengths, licenses and hardware requirements."
resource: "https://www.contextstudios.ai/guides/best-open-weight-llms-2026"
language: en
timestamp: "2026-07-27T00:00:00.000Z"
---

# Best Open-Weight LLMs 2026: Self-Hosting Guide

## TL;DR

The best open-weight LLMs in 2026 are Qwen3.5 (widest size ladder, Apache 2.0), GLM-5.2 (strongest self-hostable long-horizon and coding model, MIT, 1M context), OpenAI's gpt-oss-120b/20b (best capability-per-GPU: 117B on a single 80GB card, 21B in 16GB), DeepSeek-V4 (1.6T/49B-active frontier reasoning at 1M context), Kimi K2.7 Code (1T/32B coding agent), Gemma 4, Mistral Small 4 and MiniMax M3. Choose by hardware fit, context length and how permissive the license really is.