Tag: #deepseek-v4-flash

DeepSeek V4 Flash 0731 model introduction showing MoE architecture and agent benchmarks

Model Introduction August 5, 2026

DeepSeek V4 Flash 0731: $0.28 Beats Its Own Flagship

DeepSeek V4 Flash 0731 exits preview with 82.7 Terminal-Bench and 54.4 DeepSWE. 284B/13B MoE, 1M context, MIT license, at $0.14/$0.28 per million tokens. Agent benchmarks beat V4-Pro-Preview.

#deepseek #deepseek-v4-flash #open-source-llm #moe #ai-agent #coding-agent