❌Won't fitfp16 · 4K ctx · batch 1

Does RTX 4090 run Qwen2.5-72B? No at fp16.

At fp16 precision, Qwen2.5-72B (72 B parameters) needs roughly 146.7 GB of VRAM. RTX 4090 only has 24 GB, short by 122.7 GB. INT4 quantization brings it down to 44.0 GB.

611%
VRAM
Won't fit
Estimated VRAM (fp16)
146.7 GB
of 24 GB on RTX 4090· 122.7 GB short
Tok/sec
6
TFTT
10516ms
Power
333W
Rent on GPUniq marketplace
RTX 4090
Live prices from verified providers · billed hourly, no commitment.
Rent now

Memory breakdown (fp16)

Quantization comparison

Lower precision = less VRAM with a small quality trade-off. Quality order: FP16 > INT8 > INT4.

PrecisionVRAMUtilisationFits on RTX 4090?
FP16 (full precision)146.7 GB611%❌Won't fit
INT8 (8-bit)78.3 GB326%❌Won't fit
INT4 / Q4 (4-bit)44.0 GB184%❌Won't fit

Frequently asked

Can RTX 4090 run Qwen2.5-72B?
Not at fp16 — Qwen2.5-72B needs about 146.7 GB while RTX 4090 has 24 GB. It fits at INT4 quantization (44.0 GB) with some accuracy trade-off.
How much VRAM does Qwen2.5-72B use?
About 146.7 GB at fp16, 44.0 GB at INT4 (for a 4K context, batch size 1). Longer contexts add to KV cache size; larger batches increase activations.
What's the fastest way to run Qwen2.5-72B on RTX 4090?
Use a production inference engine like vLLM, SGLang, or TensorRT-LLM with paged attention — they cut KV cache 2–3× vs the naive estimate and batch multiple requests efficiently. For RTX 4090, enable Flash Attention 2.
Where can I rent a RTX 4090?
GPUniq aggregates live RTX 4090 offers from verified providers. You can deploy an instance in about a minute and pay hourly with no commitment.

Other GPUs for Qwen2.5-72B

Other models on RTX 4090

Try it live

Chat with Qwen2.5-72B — no setup

Send a prompt and see how Qwen2.5-72B responds directly in our chat. No installation, no GPU required to test.

Open in chat
Want to tweak sequence length, batch size, or fine-tuning? Open the full calculator →
None of the common quantization levels fit this model on RTX 4090. Consider multi-GPU deployment or a larger card.