Chinese startup DeepSeek’s V4-Flash model is the least expensive to run on benchmark tests among well-known models globally, with an average cost of 3 cents per test, according to San Francisco-based research firm Artificial Analysis. That compares with 86 cents for Moonshot AI’s Kimi K3, $1.86 for OpenAI’s GPT-5.6 Sol and $3.15 for Anthropic’s Claude Fable 5. DeepSeek’s V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens. The model, officially released last week, scored 50 out of 100 on Artificial Analysis’s Intelligence Index, which combines results from nine benchmarks. That matches Google’s Gemini 3.6 Flash and is one point behind Meta’s Muse Spark 1.1 and Z.AI’s GLM-5.2. Moonshot’s Kimi K3 scored 57, while Anthropic’s Claude Opus 5, Fable 5 and OpenAI GPT-5.6 scored nine or more points higher. DeepSeek, which sources have said is preparing for a potential IPO, is also developing a more powerful version called V4-Pro, with no release date given. Separately, Alibaba unveiled its largest AI model to date, the Qwen3.8-Max, which is not far behind in size compared with an offering