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

DeepSeek-R1-Distill-Qwen-32B vs DeepSeek-R1-Distill-Llama-70B

Neither has a clear lead on the index. DeepSeek-R1-Distill-Qwen-32B is the smaller download at 20 GB, so it runs on cheaper hardware.

DeepSeek-R1-Distill-Qwen-32BDeepSeek-R1-Distill-Llama-70B
Intelligence index88
ClassBelow every hosted tierBelow every hosted tier
Weights20 GB43 GB
QuantisationQ4_K_MQ4_K_M
Parameters32.8B70.6B
Max context128k128k
API price per 1M$0.8 in / $0.8 out$0.8 in / $0.8 out
LicenceMITMIT
Cheapest machine that runs itStrix Halo Framework Desktop, 64GB $1,959Strix Halo Framework Desktop, 128GB $3,449
Summarising usable usable
Translation usable usable
Everyday coding usable usable
Reasoning & maths good good
Agentic work don’t don’t

Ratings are coarse on purpose. Speeds and pay-back depend on the machine — open either model's page for the full list, or see both against the frontier. Context is 32k throughout.