What GPU do I need to run mlabonne/Qwen3-30B-A3B-abliterated?
30.5B parameters, published in F32. View on Hugging Face
Set up Qwen3-30B-A3B-abliteratedQwen3-30B-A3B-abliterated is published by mlabonne on Hugging Face, with 451,021 downloads and 44 likes to date. It's a Qwen3MoeForCausalLM model built for text-generation, published natively in F32.
VRAM required & cheapest live GPU fit
Required VRAM = weight size at each precision, plus a fixed overhead for KV-cache, activations, and fragmentation. Full formula and assumptions: methodology.
A GPU is only matched to a row if its hardware supports that precision, and the primary recommendation is always a single-GPU fit when one exists.
INT4 requires a quantized checkpoint actually published for this model, check its Hugging Face page before relying on this row.
Cheapest way to run Qwen3-30B-A3B-abliterated at its published (F32) precision: 3× RTX 8000 on akash, at $0.221/hr per GPU ($0.661/hr total). Quantizing to FP8 or INT4 (rows above) can cost less, but requires a compatible quantized checkpoint to exist for this model.
Qwen3-30B-A3B-abliterated: common questions
Can Qwen3-30B-A3B-abliterated run on a single GPU?
No. At FP32 it needs 136.5 GB of VRAM, more than a single 32 GB desktop card holds. The cheapest capable card in the live feed is a 48.0 GB RTX 8000, and it takes 3 of them.
Can Qwen3-30B-A3B-abliterated run in 16-bit instead of FP32?
Yes. Its published weights are FP32, 113.7 GB, or 136.5 GB once inference overhead is added. Loading the same weights in 16 bits halves that to 56.9 GB, or 68.2 GB with overhead. How much accuracy the cast costs is model-specific and is not measured here.
How many GPUs do I need to run Qwen3-30B-A3B-abliterated?
3 at FP32. It needs 136.5 GB of VRAM and the cheapest capable live offer is a 48.0 GB RTX 8000 on akash, so 3 of them come to $0.661/hr in total.
What is the least VRAM Qwen3-30B-A3B-abliterated can run in?
17.1 GB, at INT4 (quantized), which fits a 24 GB card, against 136.5 GB at FP32. That row assumes an INT4 checkpoint has actually been published for this model, so check its Hugging Face page before planning around it.
Weight-to-VRAM math, the fit rules, and how live prices are normalized: full methodology.
More mlabonne models
- NeuralMonarch-7B (7.2B, F16)
- NeuralDaredevil-8B-abliterated (8.0B, F16)
- Meta-Llama-3.1-8B-Instruct-abliterated (8.0B, BF16)
- bert-base-uncased (110M, F32)
- Qwen3-0.6B (752M, BF16)
- gpt2 (137M, F32)
