Skip to content
AI News HubLIVE
More
Summary only

DeepSeek V4 Pro (1.6T-A49B) and Flash (284B-A13B) — Base and Instruct, Runnable on Huawei Ascend

After months of delay, DeepSeek released the highly anticipated DSV4 series including Pro and Flash variants with 1M context, mixed-precision quantization, MIT license, and Huawei Ascend support. The series tops open-weight models but lags behind closed frontier models.

SourceLatent Space
DeepSeek V4 Pro (1.6T-A49B) and Flash (284B-A13B) — Base and Instruct, Runnable on Huawei Ascend
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • DSV4 Pro: 1.6T total / 49B active, Flash: 284B total / 13B active, 1M context
  • New architecture with CSA and HCA dramatically reduces KV cache to 10% of V3.2
  • MIT license, Huawei Ascend compatibility, competitive pricing but high token consumption in evals

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Models