Tag Archives: Open Source AI

The Open-Weight Models You Can Actually Run Yourself: DeepSeek, Mistral, and the Self-Hosting Option

Somewhere between a hospital that can’t send patient data to an outside API and a hobbyist who just wants to see how a model thinks, there is a growing category of AI that never phones home at all. This is the second article in LIWARSE’s five-part series on the AI landscape by type of use.Continue reading “The Open-Weight Models You Can Actually Run Yourself: DeepSeek, Mistral, and the Self-Hosting Option”

Alibaba Under the LIWARSE Lens: Qwen3.7-Max and the Qwen 3.6 Family

Alibaba’s Qwen line has quietly become one of the world’s most capable open-weight families — while its newest flagship, Qwen3.7-Max, closed the door behind it. LIWARSE closes this five-part series with the one non-Western provider whose split matters most for global AI governance.

xAI Under the LIWARSE Lens: Grok 4.5 and the Open Grok Weights

xAI ships a fast-moving closed flagship, Grok 4.5, alongside open weights it releases roughly a generation behind — and a coding agent it open-sourced after a privacy incident. LIWARSE reviews both, and the incident that shaped them.

Meta Under the LIWARSE Lens: Muse Spark and the Llama 4 Family

Meta built its entire AI identity on open weights, then quietly shipped its first-ever closed, proprietary flagship. LIWARSE reads the reversal — and what it means for the open Llama 4 models left behind — through the 3 Absolute Laws.

Google Under the LIWARSE Lens: Gemini 3.1 Pro and the Gemma 4 Family

Google runs the widest split of any major lab: a closed Gemini flagship for scale and multimodality, and a Gemma family so open it reaches from data centers down to a Raspberry Pi — including a medical-domain variant. LIWARSE reviews both.

OpenAI Under the LIWARSE Lens: GPT-5.6 Sol and the gpt-oss Family

OpenAI ships two very different kinds of intelligence: GPT-5.6 Sol, a closed frontier flagship, and the gpt-oss family, open weights anyone can download and run. LIWARSE examines both through the lens of the 3 Absolute Laws and the guarded-versus-unguarded standard.

Open Source AI Without Guardrails: The Hazard

An open model with no embedded safeguards cannot be recalled, patched, or shut off once it leaves the lab. LIWARSE argues the real fault line is not open versus closed — it is guarded versus unguarded. A five-point framework for keeping AI open to people without leaving it open to harm.