The best GPU for running Qwen3 32B (2026)
Ranked from live rental prices and computed VRAM fit, not opinion. Every card is judged at Q4_K_M — the quantisation most people actually run — at 8k context, so it's a fair comparison. A card that can't fit Qwen3 32Bat that quality isn't listed, because it could only run a crushed version.
Best value
NVIDIA GeForce RTX 5090
Most tokens/sec per rental dollar — ~68 tok/s at $0.058/hr.
Cheapest that runs it
NVIDIA GeForce RTX 5090
Lowest hourly rental that fits it — $0.058/hr, Q4_K_M.
No compromise
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition
Fastest that fits — ~68 tok/s, 21% of its VRAM.
Every card that runs it, ranked
Best quantisation each card fits at 8k context, with the cheapest live rental and a bandwidth-derived throughput estimate. Sorted by tokens/sec per dollar.
| GPU | VRAM | VRAM used | ~tok/s | Cheapest rental | tok/s per $ |
|---|---|---|---|---|---|
| NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition | 96 GB | 21 GB21% | ~68 | $1.690 | 40 |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | 96 GB | 21 GB21% | ~68 | — | — |
| NVIDIA GeForce RTX 5090 | 32 GB | 21 GB64% | ~68 | $0.058 | 1171 |
| NVIDIA GeForce RTX 3090 Ti | 24 GB | 21 GB86% | ~38 | $0.077 | 494 |
| NVIDIA RTX A5000 | 24 GB | 21 GB86% | ~29 | $0.074 | 391 |
| NVIDIA GeForce RTX 4090 | 24 GB | 21 GB86% | ~38 | $0.134 | 283 |
| NVIDIA RTX A6000 | 48 GB | 21 GB43% | ~29 | $0.267 | 109 |
| NVIDIA RTX 6000 Ada Generation | 48 GB | 21 GB43% | ~36 | $0.385 | 94 |
| AMD Radeon PRO W7900 | 48 GB | 21 GB43% | ~33 | — | — |
| NVIDIA RTX 5000 Ada Generation | 32 GB | 21 GB64% | ~22 | $0.374 | 58 |
| AMD Radeon RX 7900 XTX | 24 GB | 21 GB86% | ~36 | — | — |
| NVIDIA GeForce RTX 3090 | 24 GB | 21 GB86% | ~35 | — | — |
How this ranking is made — and its limit
Fit and throughput are computed from Qwen3 32B's own configuration with the engine behind our cost-to-run page (bits-per-weight validated to 0.7% median error). Rental prices are the cheapest live figure across the providers we track, last updated Wed, 22 Jul 2026 11:20:07 GMT.
The value column ranks renting, because that's what we can price precisely. If you're buyinga card, the fit and throughput columns are exactly what you need — but we don't publish a purchase-price value ranking, because we don't have verified street prices and won't invent them. For the buy-vs-rent decision itself, see the guides below.
Tokens/sec is a bandwidth-derived estimate, not a benchmark — we don't run our own hardware tests (editorial policy).
Check it yourself
See exactly what fits, at any context length
Running Qwen3 32B: common questions
How much VRAM do you need to run Qwen3 32B?
Can an RTX 4090 run Qwen3 32B?
Can an RTX 3090 run Qwen3 32B?
Can an RTX 4060 Ti 16GB run Qwen3 32B?
What's the cheapest way to run Qwen3 32B?
Before you buy
RTX 5090 vs two RTX 3090s
Two 3090s give you 48 GB but not double the speed for chat — llama.cpp's default multi-GPU mode makes the cards take turns. What the benchmarks actually show.
Cheapest way to serve Llama 70B
"Cheapest" flips on duty cycle and concurrency — and in Europe the electricity bill alone can approach the cost of just renting. Plus the config defaults that silently break a 16k deployment.
What --n-cpu-moe actually does
The trick that makes MoE models 5× faster — except it usually makes them slower, and the famous speedup only happens when the model didn't fit in the first place.