DeepSeek V4 Flash: Enhanced Agentic Capabilities and Strong Value

DeepSeek released DeepSeek-V4-Flash-0731, the latest in its V4 family, boasting substantially enhanced agentic capabilities. The model has 304 billion parameters and weighs 167GB on Hugging Face. Artificial Analysis ranks it ahead of MiniMax M3, which has 428 billion parameters. Its pricing is $0.14 per million input tokens and $0.27 per million output tokens, potentially making it the best value-per-intelligence model currently available. The model performs well on the Intelligence Index versus Cost per Intelligence Index Task chart. However, initial testing via OpenRouter with default reasoning level produced a disappointing output (a 'pelican'), but increasing the reasoning effort to high yielded much better results.