Forget the idea that you need a supercomputer to run the most advanced AI.
A new model from Qwen, the 3.8 27B, is changing everything, running locally on an RTX 4060 with just 8GB of VRAM. This means that, with a video card that costs around R$1.
The surprising part: a reported Kaggle setup runs Qwen 3.8 27B on 2 free NVIDIA T4s with 120k context
FP16 KV cache, 26.6 GB total VRAM, and about 14 tokens/sec. Setup is claimed under 5 minutes with no credit card.
Qwen 3.8: 550K tokens · $1.90
Grok 4.6: 2–4M tokens · $4.20 0x Alpha: 1.2M tokens · FREE
Free GPUs 2x NVIDIA T4s & run Qwen 3.8 27B at 14 t/s with 120k context
120k context on FREE Kaggle T4s. - FP16 KV - 14 t/s - No credit card. - No expensive GPU. - Total VRAM used: only 26.6 GB across both cards. - Under 5 minutes setup. - Kaggle gives you 30 hours/week of
You have reached the end of the archive
All of qwen38