Wondering how to use Ollama in ChatGPT Desktop? The quick version is just below — Ollama v0.34.0 made this integration official on 5 September 2026, and per the official docs the setup is genuinely a two-minute job from the Ollama app’s sidebar.
Short answer:
- Ollama v0.34.0 (5 September 2026, official release notes): “Ollama models can now be used directly in ChatGPT Desktop”.
- Setup: Ollama desktop app → Apps in the sidebar → connect ChatGPT (macOS;
ollama://appsdeep link on v0.34.2+ covers macOS and Windows). - Per Ollama’s integrations docs, models run in Codex mode — the coding-agent side of ChatGPT Desktop.
- No account needed for local models: v0.34.2’s setup screen has a “continue locally” option.
- Same Apps page also connects Claude Desktop;
ollama launchcovers Claude Code, OpenCode and more.
How to Use Ollama in ChatGPT Desktop: Step by Step
Here is the confirmed path, straight from Ollama’s official docs and release notes:
- Update Ollama to v0.34.0 or later. The ChatGPT Desktop integration shipped in v0.34.0 on 5 September 2026; I would go straight to v0.34.2 (15 September), which also smooths out first-run setup.
- Open the Ollama desktop app (the documented flow is on macOS) and select Apps in the sidebar. On v0.34.2+ the
ollama://appsdeep link takes you straight there on macOS and Windows. - Connect ChatGPT from that Apps page.
- Pick your open model in ChatGPT Desktop. Per the integrations docs, the connection lets you use Ollama models in Codex mode — the agentic coding side of ChatGPT Desktop.
One honesty note, because this site runs on primary sources: the release notes and docs confirm the flow above but do not document every click inside ChatGPT Desktop itself, and OpenAI’s help pages were not reachable when I checked. Expect the model picker to surface your connected Ollama models in Codex mode; if your build differs, the Apps page in Ollama is still the place everything starts.
What Shipped in Ollama v0.34.0 (and Since)
The v0.34.0 release notes carry three headline items: ChatGPT Desktop support (“use Ollama models in ChatGPT Desktop, so you can keep your existing workflow while running open models”), improved structured output performance on Apple Silicon, and OpenAI-compatible client tool search and response compaction.
Two releases later the integration story got smoother: v0.34.2 (15 September) added the first-run setup that syncs with the desktop app on macOS and Windows, plus the ollama://apps deep link. And the direction of travel is clear from the wider Apps roster — Claude Desktop connects from the same sidebar, and ollama launch fires up Claude Code, OpenCode, DeepSeek Harness, OpenClaw or Hermes Agent against your local models.
🔥 Want this set up without the guesswork? If you want local open models doing real agentic work — coding, content, SEO automation — this hybrid setup is exactly what we build. Inside the AI Profit Boardroom (3,700+ members, four live calls a week, daily tutorials, done-for-you templates and a 30-day roadmap) we build systems like this together, step by step. And if you would rather talk it through one on one first, book a free SEO strategy session and I will map it out with you.
Why Run Local Models Inside ChatGPT Desktop?
Three practical reasons I care about this integration:
- Cost control. Open models running on your own hardware do not meter tokens. For high-volume grunt work — drafts, extraction, classification — local models are effectively free after the hardware.
- Privacy. Code and content that never leaves your machine is the easiest compliance conversation you will ever have. Pair this with the v0.34.2 “continue locally” option and you have a genuinely offline-capable lane.
- One workflow. The release notes’ own framing is the selling point: keep your existing workflow while running open models. You stop choosing between the polished app and the open ecosystem — you get both in one window.
Limitations to Know Before You Switch
A practical sizing tip before you commit: match the model to the job, not to your ego. A smaller coding model that responds instantly and fits comfortably in memory will beat a giant model that swaps and stalls, especially for the repetitive agentic loops Codex mode runs. Start small, measure the failure rate on your actual tasks, and only scale up the model when the small one demonstrably cannot do the work.
Be realistic about what this is. The integration is scoped to Codex mode per the docs, so this is about coding-agent workloads, not swapping the model behind ordinary ChatGPT conversations. The documented setup flow lives in the macOS app (the deep link also covers Windows). And an open 20–70B model on a laptop is not a frontier model — it is a workhorse. Route the grunt work locally, keep frontier models for the reasoning-heavy calls, and the economics get very good.
The Bottom Line on How to Use Ollama in ChatGPT Desktop
How to use Ollama in ChatGPT Desktop, in one paragraph: update to Ollama v0.34.0+ (the feature shipped 5 September 2026, per the official release notes), open the Ollama app, hit Apps in the sidebar — or the ollama://apps deep link on v0.34.2+ — connect ChatGPT, and use your local open models in Codex mode. Two minutes of setup buys you a hybrid stack: frontier models when you need the horsepower, free local models for everything else.
FAQ: how to use ollama in chatgpt desktop
Which Ollama version added ChatGPT Desktop support?
v0.34.0, released 5 September 2026. The official release notes say you can now “use Ollama models in ChatGPT Desktop, so you can keep your existing workflow while running open models”.
Where do I connect ChatGPT Desktop to Ollama?
In the Ollama desktop app: open the app on macOS, select Apps in the sidebar, and connect ChatGPT. Since v0.34.2 the ollama://apps deep link (macOS and Windows) jumps straight to that page.
What mode do Ollama models run in inside ChatGPT Desktop?
Per Ollama’s official integrations docs, the ChatGPT Desktop integration is for using Ollama models in Codex mode — the coding-agent side of the app — rather than replacing the standard ChatGPT chat models.
Do I need an Ollama account for this?
The v0.34.2 first-run setup offers sign-in or continue locally. The integration is configured through the desktop app’s Apps page; local model execution itself does not require an account.
Which models should I try first?
Ollama’s model library at ollama.com/search tags models for coding, vision, embeddings and reasoning. Pick a coding-tagged open model that fits your RAM, and test it on a small task before trusting it with real work.
Can I connect other apps the same way?
Yes — the same Apps page connects Claude Desktop (“use local and cloud Ollama models”, per the docs), and ollama launch starts coding agents like Claude Code, OpenCode and DeepSeek Harness, plus assistants like OpenClaw and Hermes Agent.
Related reading
About the author: Julian Goldie is an SEO agency owner with 394K+ YouTube subscribers, a 100% Upwork job-success score, 75K+ community members across his groups, 10+ years in SEO and a best-selling SEO book. He runs the AI Profit Boardroom community, and you can book a free SEO strategy session with his team any time. For agency work, book a call for a custom quote.
Where to go from here: if you want local open models wired into your daily AI apps working inside a real AI SEO system rather than as a one-off experiment, join us in the AI Profit Boardroom — or book a free SEO strategy session and we will look at your setup together.
Last updated September 2026. This is the living guide to how to use ollama in chatgpt desktop — it gets updated as the tools change.
