New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Who Offers Out of the Box LLM VPS Hosting?
I am thinking about putting together a page on LowEndBox listing VPS hosting providers that directly market out of the box ready instances or containers for running open weight LLMs, similar to our Cheap GPU List:
https://lowendbox.com/cheap-gpu-list-nvidia-gpus-for-ai-training-llm-models-and-more/
GPU and CPU based services are both relevant. I am specifically looking for products designed and marketed for easy LLM deployment, ideally with tools like Ollama, llama.cpp, vLLM or some of the newer Asian models, already installed or available through a template.
Who in the LEB/LET ecosystem is offering this today?
Thanked by 1buggedout
Comments
@Andreix soon™
@nube (seems like reselling)
Hetzner
No one else?
Was hoping to see more providers from the Leb ecosystem jumping on this train.
Hosted LLM? Not interested, too much customer personal data to go wrong, i see at work already that people will put anything in there asking for anything with zero shame or thought, even logged into an account.
Now hardware and some model preinstalled, sure, but that just needs an image im sure exists already and enough RAM
Hi,
we are talking here about a VM / container that has "just" preinstalled openclaw or similar personal assistants where customers will feed their AI accessdata to feed their token?
Or are we talking about VM's that will provide multiple GB of VGA RAM and CPU Cores that free/local LLM's like gpt-oss, Qwen, Gemma, DeepSeek, ... will be self hosted?
The 1st can be easy and will also be more or less compatible with what people might search at low endtalk.
The 2nd is something where things will start with multiple hundreds or even thousands of USD/EUR per month - maybe not the typical low endtalk ;-)
OVH
The latter but with easy deployment. Customer rents an instance with ideally some GPU, ollama/similar gets installed automatically.
It doesn't have to cost hundreds or thousands per mo, it just depends on the size of the model. You can easily run decently sized models on mid-tier RTX 3000 series
I hope any host who attempt to enter the market understand that the current floor (low end model, set by openai and deepseek) is likely to receive another 90% discount by Feb 2027 as NVIDIA's Vera Rubin went online.
This isn't the end of logarithmic token price reduction because there's more novel memory solution like Z-angle, HBF(flash), XBM, etc on the horizon (for Y28)
This is lowendtalk, really think offering LLM-focused GPU VPS will fit in LET rules
Personally I think it's better to host a small model locally, especially if you have a 16-32GB VRAM GPU or a Mac Mini with unified memory (there's also some sweet newer Ryzen/Intel mini-PCs with unified memory.)