DeepSeek-V4-Pro Weights Released with Million-Token Context

cedric_chee · x · 2026-08-13

The weights for the DeepSeek-V4-Pro model have been officially released on Hugging Face. According to the repository details, the model focuses on highly efficient million-token context intelligence and supports 8-bit and fp8 precision under the MIT license.

Related event: DeepSeek-V4-Pro Officially Released with Enhanced Agent Capabilities and Peak Pricing(48 posts)→

Original post →

More from Models

Models channel →