Can PewDiePie’s AI Actually Replace Claude?

Odysseus workspace is free, but Ajax model isn't out yet, here's what you can actually run today on your hardware and the real privacy catch. PewDiePie's announcement split into two parts: Odysseus, a free self-hosted AI workspace you can download now, and Ajax, a fine-tuned Qwen 3.5 9B model still in development. The workspace connects to existing local models via Ollama, remote APIs, or cloud endpoints, so you don't need to wait, test it today using inference you already have running. The real cost isn't hidden subscription fees; it's the compute, configuration, and operational overhead you absorb yourself, whether running models locally on your GPU or CPU, or paying external API providers for tokens. Before buying new hardware, measure your actual memory needs. A nine-billion-parameter model at four bits per weight is roughly 4.5 GB raw, but once loaded into an inference engine like Ollama with context windows and working memory for tool calls, cache, and document processing, real-world usage demands far more. Use ollama ps to inspect how your current setup handles your everyday tasks, whether the model runs on GPU, CPU, or split between both. If your existing machine keeps up, skip the purchase and test the interface directly. Practical workflows worth trying: extract deadlines and draft replies from a document without sending email, then verify the output matches source dates; run web research and manually check cited links for accuracy; test tool calling and endpoint settings separately from conversational smooth talk. The Odysseus threat model documents privileged admin access to shell, files, and email alongside a default approval gate for risky operations, but that gate isn't process isolation, it's a confirmation pause. Start with bounded non-sensitive trials on a non-admin account, keep the approval gate on, and verify your endpoint routing before granting deeper permissions. This video is for anyone considering local AI setup, curious about the workspace itself, or wondering whether an unreleased fine-tune justifies new hardware spending. Chapters: 0:00 Intro 1:10 Start Here 2:32 Your Computer 5:05 Useful Work 7:40 Private Where? 10:12 The Verdict Tools & resources mentioned: - Odysseus: https://github.com/odysseus-dev/odysseus - Ollama: https://ollama.com - Qwen 3.5: https://ollama.com/library/qwen3.5 - LM Studio - llama.cpp - vLLM - SearXNG About The Stack The Stack helps you build with AI. Each video takes one tool, model, or workflow and shows how it works in a few focused minutes, with the real benchmarks and real costs. We go deep on Claude Code and Cursor for AI coding, AI agents and MCP servers, the open-source AI tools and GitHub repos most people miss, RAG and vector search, fine-tuning, and running local LLMs on your own machine with Ollama and LM Studio. We compare models like ChatGPT and Claude, test AI automation with Zapier, Make, and n8n, and flag the tools that actually ship. Subscribe for new breakdowns: https://www.youtube.com/@the-stack-ai?sub_confirmation=1 #odysseus #LocalAI #free ai models