New Deepseep’s new pattern is to run on a single GPU

DEepseek’s Updated R1 Reasoning AI could be obtained the bulk of the community attention to this week. But the Chinese Lacket, librilled the “distilled” version of his new R1, DEPENSEEK-R1-052-QWEN3-8B ​​Models completely for certifications

The smaller updated r1, which was built using the Qwen3-8b model Alibaba launched in May as a foundation, execute better than Google Gemini 2.5 flash on the Aime 2025, a collection of challenging math questions.

Deepseek-r1-0528-qwen3-8b also corresponds to microsoft recently Phi 4 Reasoning Plus model on another test of math skills, hmmt.

The distilled patterns as deepseek-r1-0528-qwen3-8b are generally less able than their fullest companions. On the side of the more, I am very meno showed. According to To the cloud Nodeshift, Qwen3-8b needs a GPU with 40GB-80GB of RAM to run (eg a nvidia h100). The needs of the new r1 r1 r1 Around a dozen 80gb GPUS. I am

DEEPSEK has formed Deepseek-R1-0528-Qwen3-8b by taking the text generated by the R1 update and using in LOT QWEN3-8B. On a patterned web page on the pattern on dev plazing fafare deepseek-r1wen3-8b by the ragon and conducted patterns on the patterns of the small scale. ‘

DEepseek-r1-052-qwen3-8b is available under a permitted license of mit, which means it can be used non-restriction. Many guests, including Lm studioalready offered the model across a API.