DeepSeek V4 Flash is narrowing the gap with Anthropic's Claude through lower API costs and a context of 1 million tokens. However, it remains challenging to surpass the Claude model in public benchmarks. The V4 Flash uses 284 billion parameters, with the active model being 'DeepSeek-V4-Flash-0731'. The pricing is set at $0.66 off-peak and $1.32 peak, which is cheaper compared to Claude's output fee of $25 for the Opus 4.8. However, DeepSeek plans to raise the API prices for V4 Pro and V4 Flash, implementing a new pricing plan starting August 17. In performance comparisons, V4 Flash scored 82.7 on the Terminal Bench 2.1, while Claude scored 85.0, indicating a small difference. Nevertheless, Claude showed superiority in tests like NL2Repo and CyberGym. DeepSeek emphasizes that V4 Flash has expanded the cost-performance range, allowing enterprises to choose models based on top performance and bulk call tasks. The cost-saving effects of V4 Flash for enterprise AI need to be validated through actual implementation cases.
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.

















