My AI stack

Jun 19, 2026 #ai #updates

This article is a living document - captures a snapshot of my current toolset, and occasionally updates on why I changed it.

AI chat
- grok 4.5 chat at grok.com (use “Expert” to ensure grok 4.5 or latest)
- huggingchat with inference providers, or run huggingchat locally with local models
Primary terminal AI harness (for anything involving local file system, coding or otherwise): Pi
Hosted AI provider: openrouter
Local AI provider: ollama
AI model: (* indicates no vision/image input)
- simplest to hardest tasks
  - Qwen3.6 27B
  - Deepseek-v4-Flash *
  - Minimax M3
  - GLM 5.2 *
  - Grok 4.5
  - Kimi K3
- Alternates
  - Nemotron 3 Ultra *
  - Gemma 4 (for local)
NotebookLM - For deep research and different “views”
Grok - For the pulse of the internet AI chats (with X data) and also for a frank uncensored response on many things.
Gemini - For web index data and Google properties
Hermes agent and OpenClaw, but always in a VM - High agency agents
Asta - Ai2. Literature-discovery agent over Semantic Scholar.

Software stack (highly optimized for AI):

OS - Mac
Mac VM in Mac - Lume (from cua)
Terminal - iTerm2
IDE - VSCodium (with telemetry off)
Python - uv, pydantic v2, FastAPI, basedpyright
data - Files - json, markdown with file system conventions, or if needed, DB - sqlite
Native app - Pyside 6 for Qt
Web frontend - vanilla html/css/js
Obsidian - Personal vault, daily notes, frameworks.

Past luminaries

Claude - As of today (2026-06-19), Claude and Claude Code are a mess - since Opus 4.8, the entire experience with Claude and Claude Code went south, then there was the Fable debacle, and even before that the “no subscription access, even at 200$ per month to best model” - the sota model edge is already being held by a sliver. I think they just gave it away with all the recent choices - both technical and business. Still use it rarely when other things dont work, but there are better options for most things.
ChatGPT - Was a favorite for a long time. They went off alignment several months back - heard they are better now, am checking them out, but no reason really to go back - cheaper models are as good if not better for most use cases.