Open Source

How much will it cost to host something like qwen3.6 35b a3b in a cloud?

As hardware evolves, devs ask: how much to rent cloud GPUs for Qwen 3.6 35B?

Deep Dive

A Reddit user asks how much it will cost to host the Qwen 3.6 35B A3B model for coding until they upgrade their hardware.

Key Points
  • Qwen 3.6 35B A3B requires 10–20GB VRAM depending on quantization, pushing most users to cloud GPUs.
  • Estimated cloud costs: $0.20–$1.60/hour on platforms like Vast.ai or RunPod using A100 or H100 GPUs.
  • For continuous use (24/7), monthly costs range from $300–$600; per-token serverless APIs are cheaper for sporadic use.

Why It Matters

Affordable cloud hosting for 35B MoE models opens coding AI to users without high-end GPUs.