@future.with.ai98: China just made OpenClaw free forever 🤯 Install Qwen 3.6 through Ollama and you can run OpenClaw locally with zero API fees. Qwen 3.6 is an absolute coding monster, and when you combine it with OpenClaw you get a powerful AI agent setup that costs nothing to use. This might be the best free OpenClaw setup yet. #Qwen36 #OpenClaw #ChinaAI #AI #ArtificialIntelligence

Luke :

still need to pay for Ollama running openclaw

ForwardCompatible :

qwen 3.6 pro is not free forever. it's free while ollama pays them for it. there are local Qwen3.6 versions you can use, and they do kick butt, but 3.6 pro is their coding and reasoning "monster" And it is most certainly not local inference capable at almost any quant.

JiggedyJig :

FFS... the models you are able to run locally is crap.

xterminatorlite :

Bro my gpu is melting

Randumb :

good luck running this on any consumer hardware. You need a beast of a machine just to get a basic functionality working

STAR LORD :

you need hardware to run a good model. at least 5000$ investment. Soooo... give me a break.

Besian Sherifaj :

Its not better than 3.5 27b based on my personal experience! On 4090 !

optimistic🫆🇰🇪🇩🇪🇬🇧 :

can I run it on lenovo thinkpad

Pavel Ponomaryov :

If something is free and isn't hosted on your server, then you are the product.

canyonero :

how does it compare to gemma4?

Patryk Chowratowicz :

what is minimal gpu i need?

R.J. :

except ollama is not free

Anonymous Privacy :

ohh yeah?? Free? they also provide the GPU? you know you need one right? and not just a basic gaming one but a monster with enough VRAM to run large models

pkollitsch :

any model you run locally "removes API fees forever".

juli :

But do you need all that RAM and a good GPU in general for it to work or is that only for bigger projects? cause i‘m vibecoding a small webapp on my laptop with a 3060 and 16GB DDR5, would that be enough or shouldn’t I even bother to try? (Same question with claude code etc, but I‘m looking for a free alternative for now)

AmerZara :

you forgot to mention it need minimum 24gb vram to run

Cristiano Armani :

Fire 🔥 :

sure with 24gb of vram

Stefan Alt :

I am absolutely with you. I do it already on my local ai system. You can do it with gamma too. But not everyone have a system where they can run usefully big models

Techtonic reviews 🤩 :

It's not good for agentic workflows

Mił :

Mine 5070 ti 16gb is crying

AD :

link to download

Gezzamondo 🏴󠁧󠁢󠁳󠁣󠁴󠁿 :

You need to start telling the full story and not just some of it making it sound like you get full models for free

nicoltms4r1🇨🇦 :

Not many people have sufficient VRAM on their hardware to run this locally. Need over 20GB for meaningful results.

cloudynights :

china we love you