OpenLLM - Take your AI tokens anywhere

Hey Product Hunt, I'm Jonathan, co-founder of OpenLLM (openllm.sh).

Take your tokens anywhere. Use your subscriptions, API keys and local models in Codex and the coding agents you already use, and keep working when you hit a limit.

You probably already pay for some AI subscriptions and API keys, and maybe run a local model or two. Each one is stuck in its own app. I was paying for a lot of intelligence and could only use it in the box it came in. With OpenLLM you connect them once and get one key, and your coding agents all use it.

Here's what comes with it:

- Remote sessions. Start a coding session on your home machine from your laptop, in the web app or the Mac app. If the machine is online, it just opens.

- Fallback. Put your models in order, and when you hit a limit the next one picks up.

- One memory across all your models and apps. Save a decision or preference once and every app you connect can recall it. It's on every plan.

- The ChatGPT app. On a Mac, run the ChatGPT app on your OpenLLM models, so you can use your subscriptions inside an app you already know.

- Free inference, DeepSeek and Qwen, so your first request works before you connect anything.

The same key also works in any app that runs on AI, anywhere an OpenAI or Anthropic key does.

You can connect Claude Code, Codex, Kimi Code, Grok, Cursor and Muse Code, plus API keys, custom API keys, local models and a Brave Search key. You can also see which models answered and how many tokens you used, and there are free context.dev credits so your agents can read the web. An MCP marketplace is coming soon.

Free to start. Pro is $30/mo for up to 5 devices, with a 3-day trial. Works on macOS and Linux.

For Product Hunt: code PHUNT gets 50% off Pro for 6 months, for the first 100 people.

Teams is coming soon: set token limits per person, see which apps your team runs, give everyone the same tools and MCPs, and share one memory across all of them.

It doesn't make any subscription unlimited, and it isn't account sharing. Your vendor's limits and terms still apply, and it's for your own accounts on your own machines. The question I get most is "will this get my account banned?", so I answered it in the comment below.

Docs: https://docs.openllm.sh

I'll be here all day. Tell me where you want to use your tokens and I'll tell you how, or fix it if it breaks.

Good to see which model answered is visible. When a fallback happens halfway through a coding session, I'd want to know where that switch occurred.

What subscriptions can I add?

The fallback chain is the part I'm most curious about, especially with a local model in it. If a session starts on a hosted model and falls back to a local one mid-task, does the local model get the same shared memory and enough of the conversation to pick up where the other left off, or does it effectively restart with a summary? And can a local model sit first in the chain, so the hosted model only steps in when the local one hits a limit?

the shared memory is what got me here. if I save a decision from claude code, does codex pick it up mid task or only when I start a new session?