Home / Models & Local AI

Models & Local AI

Choose models for your tasks and hardware. Check weights, licensing, quantization, context and inference interfaces.

New analysis and existing guides, ordered by date. Check each article’s versions and testing limits; inclusion here does not mean an older guide was retested.

2026-10-29

Choosing Qwen Locally: Separate Qwen3.8-27B Weights, Qwen Code and Inference Runtimes

Choose Qwen3.8-27B weights, Qwen Code and inference runtimes separately. Check licenses, quantization, context and hardware without invented benchmarks.

Updated 2026-02-28

Enterprise LLM Benchmark 2026: GPT-5.3 vs Claude 4.6 (Fully Rewritten)

A task-level enterprise benchmark for GPT-5.3 and Claude 4.6 across engineering, long-form knowledge work, and support automation—with routing and rollout guidance.

2026-02-24

Build a Local AI Agent with OpenClaw on Raspberry Pi (2026 Guide)

Step-by-step guide to running OpenClaw on a Raspberry Pi 5 with Ollama for a private, always-on AI assistant. Covers hardware, installation, Home Assistant integration, and Tailscale remote access.

2026-02-21

OpenClaw + Ollama Local Offline Deployment Complete Guide (2026)

Zero API cost, fully offline AI assistant! Complete walkthrough of deploying Ollama locally, configuring LLaMA, Qwen, DeepSeek models, and seamlessly integrating with OpenClaw for a complete local AI agent system.

Updated 2026-02-16

OpenClaw Model Fallback Chain Guide: Rate Limits, Failover & Cost Optimization

How to configure cross-provider model fallback chains in OpenClaw to handle 429 rate limits, provider outages, and cost control. Full config examples included.

2026-02-10

Best AI Models for OpenClaw in 2026: Practical Comparison

Compare GPT, Claude, Gemini, and local models for OpenClaw. Covers cost, speed, reasoning quality, and which model fits each task.