DeepSeek Releases V4.1 Flash Multimodal MoE and Begins Phased Retirement of V4 Pro
Releasing a frontier-scale MoE model under the MIT license with drastic KV cache compression sharply reduces enterprise inference costs for large-context workflows.
Archived edition · Bangkok time
This edition covers DeepSeek's release of the 552B multimodal MoE V4.1 Flash, the resignation of an Anthropic researcher raising superintelligence safety concerns, and Apple's announcement of the foldable iPhone Duo powered by 2nm A20 Pro silicon.
3 unique stories
Releasing a frontier-scale MoE model under the MIT license with drastic KV cache compression sharply reduces enterprise inference costs for large-context workflows.
The departure of a core capabilities researcher signals that existential concerns over recursive self-improvement are no longer theoretical policy debates, but technical realities felt inside the engineering teams building frontier foundational systems.
Apple's transition to a commercial foldable form factor paired with 2nm silicon cements on-device multimodal edge computing as a primary enterprise standard, challenging mobile software architectures and IT budgeting across global and regional enterprises.