Local AI's Resurrection: Reasoning Models and Workflow Automation

Tools

Ollama 0.9 adds model resurrection for seamless long sessions

Ollama's new model resurrection feature is a huge quality-of-life improvement. It automatically restarts models in the background when your session history gets too long, which should eliminate a lot of the manual tinkering and 'context window full' errors that plague long coding sessions.

Avante.nvim supercharges local AI coding in Neovim

The Avante.nvim plugin for Neovim just got a major update aimed squarely at local model users. It now supports Ollama, LM Studio, and other local backends natively, giving developers a fast, modal workflow for AI-assisted coding without leaving their editor.

News

Alibaba's QwQ-32B model rivals DeepSeek R1 reasoning in a smaller package

The new QwQ-32B model is a direct challenge to R1, offering reasoning capabilities in a much smaller, more efficient package. For local inference, a 32B parameter model that can reason is a massive win for deployability on consumer GPUs compared to the larger R1.

Analysis

A tactical guide to building a local LLM codegen workflow

Harper's practical walkthrough of his LLM codegen workflow is a goldmine for builders. He breaks down his exact setup using local models for the heavy lifting, showing how to avoid the hype and build a reliable, automatable pipeline for daily development work.