DeepSeek V4 Flash Free

According to an August 12, 2026 report citing Artificial Analysis benchmark data, DeepSeek V4 Flash held the lowest cost per task among the top 20 models as of August 7, at roughly 3 cents per task. The article situates V4 Flash alongside other recent Chinese releases such as Moonshot AI's Kimi K3 and Alibaba's Qwen 3. The piece frames the result within a broader enterprise-AX spending context, referencing Uber exhausting its 2026 AI budget by April due to Claude Code usage across ~5,000 engineers and Amazon's internal Claude Sonnet project ballooning 860% over budget to roughly $1.8 million. Against that backdrop, DeepSeek V4 Flash'

DeepSeek released DeepSeek-V4-Flash-0731 on July 31, 2026, as the official release of the V4 Flash efficiency-tier language model, superseding the April preview. The version stamp "0731" indicates the July 31 re-release, not a different product; the API model ID remains "deepseek-v4-flash" and the architecture and para DeepSeek V4 Flash is a mixture-of-experts language model with 284 billion total parameters and 13 billion activated per token, supporting up to one million tokens of context and up to 384K output tokens. As of the 0731 release, official pricing is $0.0028 per million cache-hit input tokens, $0.14 per million cache-miss

DeepSeek V4-Flash supports both thinking and non-thinking modes, allowing developers to choose between faster responses and deeper reasoning. It also supports a 1-million-token context window and up to 384K output tokens, combining long-context reasoning with agentic and coding capabilities. DeepSeek describes V4-Flash DeepSeek has warned that its API prices are going up, though the company has not yet announced new prices or an exact effective date. As of the article's publication, current published API prices are $0.0028 per million cache-hit input tokens, $0.14 per million cache-miss input tokens, and $0.28 per million output toke