Sunk Cost sunkcost.ai Data checked 2026-09-03

DeepSeek-R1-Distill-Qwen-32B vs Qwen3 32B

DeepSeek-R1-Distill-Qwen-32B scores higher on the intelligence index, 8 against 7. Both take the same machine to start: the cheapest here that runs either is the Radeon AI PRO R9700, 32GB, at $1,299. The Radeon AI PRO R9700, 32GB is priced as the card alone, without the PC around it. On it, they run at much the same speed: 20 and 20 tok/s, Qwen3 32B's measured and the other estimated from memory bandwidth. At 500k tokens a day the Radeon AI PRO R9700, 32GB pays for itself in 9.4 years running DeepSeek-R1-Distill-Qwen-32B, against 150 years running Qwen3 32B.

DeepSeek-R1-Distill-Qwen-32BQwen3 32B
Intelligence index87
ClassBelow every hosted tierBelow every hosted tier
Weights20 GB20 GB
Needs at 32k28 GB28 GB
QuantisationQ4_K_MQ4_K_M
Parameters32.8B32.8B
Max context128k40k
API price per 1M$0.8 in / $0.8 outpriced as DeepSeek-R1-Distill-Llama-70B$0.08 in / $0.28 out
LicenceMITApache 2.0
Machines here that run it26 of 3726 of 37
Cheapest machine that runs itAMD Radeon AI PRO R9700, 32GB $1,299card onlyAMD Radeon AI PRO R9700, 32GB $1,299card only
Summarising usable good
Translation usable good
Everyday coding usable good
Reasoning & maths good usable
Agentic work don’t usable

Run DeepSeek-R1-Distill-Qwen-32B on the Radeon AI PRO R9700, 32GB · or Qwen3 32B

Ratings are coarse on purpose: they say what a model is usable for, not where it places to the decimal.

Side by side on the Radeon AI PRO R9700, 32GB

The cheapest machine that runs either model is the same one, so this is the pair doing the same work on the same hardware: AMD Radeon AI PRO R9700, 32GB, at $1,299card only.

DeepSeek-R1-Distill-Qwen-32BQwen3 32B
Speed at 32k20 tok/s estimated20 tok/s measured
Pay-back on this machinePays back in 9.4 yearsPays back in 150 years
API cost per month$12.18priced as DeepSeek-R1-Distill-Llama-70B$1.41

Run DeepSeek-R1-Distill-Qwen-32B on the Radeon AI PRO R9700, 32GB · or Qwen3 32B

How much use it takes to pay for the machine

Everything above is at 500k tokens a day. Pay-back moves with how much you actually run, so here are both models at the five levels of use the calculator names, on the Radeon AI PRO R9700, 32GB. DeepSeek-R1-Distill-Qwen-32B pays for it sooner at every level of use, so which of them to run does not turn on how hard you work it.

A day's useDeepSeek-R1-Distill-Qwen-32BQwen3 32B
50ka few chats a day94 years1,499 years
200klight assistant use24 years375 years
1Ma moderate coding-assistant day4.7 years75 years
4Mheavy coding with an agent14 months19 years
20Magents running most of the day2.8 months3.7 years

Run DeepSeek-R1-Distill-Qwen-32B at 20M tokens a day · or Qwen3 32B

Memory is not what separates them

DeepSeek-R1-Distill-Qwen-32B needs 28 GB of memory at 32k of context and Qwen3 32B needs 28 GB. Every machine priced here that runs one runs the other, at 32k of context. So the choice between them is what each is good at, how fast it runs and what the same work costs on an API, not what you have to buy to hold it.

DeepSeek-R1-Distill-Qwen-32B is also head to head with Devstral 2 123B above it on the leaderboard and DeepSeek-R1-Distill-Llama-70B below it. Qwen3 32B is also head to head with Llama 3.1 8B Instruct above it on the leaderboard, Mistral Small 3.2 24B Instruct below it and Qwen3.8 27B, the current Qwen nearest it in size.

The assumptions behind both columns

Both columns use the same usage: 500k tokens a day at 15:1 input to output, 32k of context, $0.17 per kWh, and today's API prices held flat. Speeds marked estimated are worked out from memory bandwidth rather than measured, and pay-back scales with them. Where nobody rents an open model by the token, its API prices are the nearest hosted model's, named beside them. Machines are the 37 here with a published price that are still sold. Change any of it in the calculator.

More head to head: every machine that runs DeepSeek-R1-Distill-Qwen-32B · every machine that runs Qwen3 32B · every other match-up · both against the frontier · the quickest pay-back at each level of use