[AINews] DeepSeek v4.1-Flash: 763B-P8B-D16B novel causal Encoder–Decoder architecture with vision marks the Return of the Whale

Latent Space · 1 month ago · 5 ● 5 sources

DeepSeek released the open-weight model DeepSeek v4.1-Flash with a causal encoder-decoder architecture and added vision inputs. It charges $0.30 per 1M input tokens and $1.20 per 1M output tokens (MIT license, 1M-token context), with active parameters listed as 8B for prefill and 16B for decode. The launch shifts attention toward inference efficiency—especially KV/cache cost reduction—along with faster, lower-RAM local deployment enabled by SSD offloading and day-0 support in tools like Ollama and Baseten.