AI Summary15 Jun 202630 Dhuʻl-Hijjah 1447 AH
AI Summary
15 Jun 202630 Dhuʻl-Hijjah 1447 AH
DeepSeek V4 and Huawei Ascend 950DT: The Co-Designed Chip That Cut AI Inference Costs by 75%
A new analysis by SemiAnalysis reveals that Huawei's AI accelerator Ascend 950DT and DeepSeek V4 were co-designed, altering the conventional understanding of model development. This collaboration has resulted in a 75% reduction in the V4-Pro API pricing, bringing the inference cost for one million tokens down to 0.20 yuan, significantly lower than Anthropic's offerings.
The study also indicates that Huawei's CANN 8.5 software suite and the unique architecture of the Ascend 950DT were developed with DeepSeek's inference patterns in mind, enhancing performance.
Follow these topics
Sign in to follow the topics that matter to you
This summary is generated with AI and receives periodic editorial review. Refer to the original source for full details.
DeepSeek V4 and Huawei Ascend 950DT: The Co-Designed Chip That Cut AI Inference Costs by 75%
A new analysis by SemiAnalysis reveals that Huawei's AI accelerator Ascend 950DT and DeepSeek V4 were co-designed, altering the conventional understanding of model development. This collaboration has resulted in a 75% reduction in the V4-Pro API pricing, bringing the inference cost for one million tokens down to 0.20 yuan, significantly lower than Anthropic's offerings.
The study also indicates that Huawei's CANN 8.5 software suite and the unique architecture of the Ascend 950DT were developed with DeepSeek's inference patterns in mind, enhancing performance.
Infrastructure
Follow these topics
Sign in to follow the topics that matter to you
This summary is generated with AI and receives periodic editorial review. Refer to the original source for full details.
0
0 reading now
Insight Score
Rate to unlock
You have not rated yet.
Sign in to react, rate, and save. Sign In
