Ajax vs Odysseus: the model vs the workspace

The single most confusing thing about this story: two Greek-myth names, two different things. Odysseus is software you can install today. Ajax is a model that hasn't been released.

Ajax vs Odysseus, side by side

AjaxOdysseus
What it isA fine-tuned Qwen 3.5 9B language model - the "brain"A free self-hosted AI workspace - the "body": chat, agents, research, documents, email, notes, calendar, local-model workflows
Released?No - official page says "when it's ready" (as of 2026-10-07)Yes - public on GitHub since May 2026
LicenseNot announcedNot confirmed yetAGPL-3.0
Where to get itdata.pewdiepie.com (only official point; nothing posted yet)github.com/odysseus-dev/odysseus - docs at odysseus-dev.github.io/odysseus
Runs other models?It is a modelYes - Odysseus existed before Ajax and runs other local models; Ajax is the first model trained specifically for it

Get Odysseus for your own computer

Start with the official Quick Start, then use the repository's Setup Guide for your operating system. The default branch is dev; the maintainers describe main as the more curated option. This downloads the workspace, not Ajax weights.

Odysseus can use local models or an API endpoint. Choose a local model endpoint for local inference; an external API still processes requests with its provider. Installing the workspace alone does not make that API local.

Official setup links checked October 21, 2026. We have not tested a fresh Odysseus installation here.

Why the names

In Greek myth, Ajax was Odysseus' rival. PewDiePie joked the model is named after the cleaning product (Tom's Hardware). Either way, the myth pairing is the point: the model is built to live inside the workspace.

How we got here

Odysseus came first. PewDiePie launched the free self-hosted AI workspace around May 2026 (some outlets say June), and by July the repo had moved to the odysseus-dev GitHub organisation. It already runs other local models - chat, agents, research, documents, email, notes and calendar workflows, all on your own hardware. Tom's Hardware compared its setup to OpenClaw and Nous Research's Hermes Agent.

Ajax arrived on 2026-10-01 in a video titled "I seriously should NOT be dropping this." It's a fine-tuned Qwen 3.5 9B model, trained specifically to be Odysseus' always-on agent - the brain the workspace was missing. A countdown on the official page pointed at 2026-10-03, then vanished; the release is now "when it's ready" with no announced date (full timeline).

Why people confuse them

Three reasons. First, both names come from the same person in the same few months - Odysseus launched in May, Ajax was unveiled in the 2026-10-01 video. Second, coverage headlines say things like "PewDiePie's Ajax AI" while showing Odysseus screenshots, because the demo runs inside the workspace. Third, you genuinely can't use Ajax-as-intended without Odysseus - the model is designed as Odysseus' agent, so the release story is really about the pair. Odysseus' setup has been compared by Tom's Hardware to OpenClaw and Nous Research's Hermes Agent.

What Ajax adds that Odysseus alone doesn't

Odysseus is model-agnostic plumbing - it can already drive other local models today. What Ajax brings is a brain trained for that specific loop: the official page describes it as "a fine-tuned Qwen 3.5 9B model, trained for Odysseus to be your always on agent," handling daily tasks "from search to browse web to email to your calendar" privately. Two other differences: its refusal has been ablated ("a freer, less restricted AI experience," per the official wording - use responsibly), and at 9B parameters it's sized for a single consumer GPU rather than a server (hardware estimates).

Quick answers

Which one do you need?

Related: hardware estimates, download status, FAQ.