DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm says

• DeepSeek V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens, making it far cheaper than competitors including OpenAI's GPT-5.6 Sol ($1.86 per test) and Kimi K3 ($0.86 per test)
• On performance benchmarks, V4-Flash scored 50 out of 100 on Artificial Analysis's Intelligence Index, matching Google's Gemini 3.6 Flash but trailing OpenAI and Anthropic models by 9+ points
• DeepSeek faces intensifying competition from Chinese rivals including Moonshot, Alibaba, and ByteDance, all targeting businesses seeking cost-effective AI deployment at scale

Summary: DeepSeek's V4-Flash Model Sets New Cost Benchmark in AI Market

Chinese AI startup DeepSeek has released its V4-Flash model, establishing itself as the lowest-cost AI option among major global models. According to research firm Artificial Analysis, V4-Flash costs approximately 3 cents per test to run—dramatically cheaper than competitors and more than 105 times less expensive than Anthropic's Claude Fable 5 ($3.15 per test).

Key Pricing Data:

• DeepSeek V4-Flash: $0.14 per million input tokens, $0.28 per million output tokens
• Competitors: Moonshot's Kimi K3 at 86 cents, OpenAI's GPT-5.6 Sol at $1.86, Claude Fable 5 at $3.15 per test

Performance Metrics:

The V4-Flash scored 50 out of 100 on Artificial Analysis's Intelligence Index, matching Google's Gemini 3.6 Flash and trailing slightly behind Meta's Muse Spark 1.1. However, it significantly lags top performers—Moonshot's Kimi K3 (57), and Anthropic's Claude and OpenAI's GPT models (59+).

Market Context:

DeepSeek is attempting to regain momentum after its R1 model caused a global tech stock selloff in early 2025 by demonstrating ultra-low-cost AI capabilities. The company faces intense competition from Chinese rivals including Moonshot, Z.AI, ByteDance, and Alibaba, all targeting businesses seeking affordable AI deployment at scale.

DeepSeek is reportedly preparing for an IPO and developing a more powerful V4-Pro version (release date TBA). Separately, Alibaba launched its Qwen3.8-Max model on Monday, further intensifying the competitive landscape.

The comparison methodology accounts for data processing requirements, providing a more realistic value assessment than headline pricing alone.