What hardware do you need to run Qwen3 32B?
Qwen3 32B at Q4_K_M is 20 GB of weights, with a context ceiling of 40k tokens. Strong for day-to-day coding help. Slips on long multi-step chains.
How good is it, really?
On the Artificial Analysis Intelligence Index v4.3 it scores 7 (reasoning mode), which puts it in the Below every hosted tier band. Fine for simple, well-specified tasks. Noticeably less capable than anything the big labs sell today. Score source. See the whole table.
- Summarising — good
- Translation — good
- Everyday coding — good
- Reasoning & maths — usable
- Agentic work — usable
What it costs either way
Renting the same model costs $0.08 per million input tokens and $0.28 per million output (OpenRouter, cheapest active endpoint, checked 2026-09-03). Buying a machine only beats that if you use it hard enough, for long enough, that the hardware price divides down below the rental bill.
Machines that run it
| Machine | Price | Speed at 32k | Pay-back | |
|---|---|---|---|---|
| Strix Halo Framework Desktop, 64GB | $1,959 | 6.8 tok/s estimated | Pays back in 311 years | Run the numbers |
| Mac mini M5 Pro, 48GB | $2,299 | 8.1 tok/s estimated | Pays back in 302 years | Run the numbers |
| Mac Studio M5 Max, 48GB | $3,099 | 16 tok/s estimated | Pays back in 257 years | Run the numbers |
| MacBook Pro M5 Pro (16-inch), 48GB | $3,599 | 8.1 tok/s estimated | Pays back in 473 years | Run the numbers |
| DGX Spark GB10 Grace Blackwell, 128GB | $4,699 | 7.5 tok/s measured | Pays back in 767 years | Run the numbers |
One machine per family, cheapest first. Speeds are measured where a public benchmark exists and estimated from memory bandwidth otherwise; the calculator says which for any configuration.
The specifics
- Parameters
- 32.8B
- Quantisation
- Q4_K_M
- Weights on disk
- 20 GB
- KV cache
- 8.6 GB at 32k context
- Maximum context
- 40k tokens (32k natively; 128k with YaRN)
- Licence
- Apache 2.0
- Sources
- source 1, source 2