DeepSeek V4 Pro is already huge

Name: DeepSeek V4 Pro is already huge
Uploaded: 2026-08-22T13:00:01.000Z
Duration: 8 min 35 s
Description: 1.6T total parameters. 49B active per token. 1M-token context. And it uses token-wise compression + sparse attention so it doesn’t need to reread everything constantly. But long agent tasks can still drift.

1.6T total parameters. 49B active per token. 1M-token context. And it uses token-wise compression + sparse attention so it doesn’t need to reread everything constantly. But long agent tasks can still drift.

Julian Goldie SEOAug 221

Keep scrolling for the next clip ↓All of deepseek Archive