Pricing just became the battleground in AI competition. DeepSeek V4-Flash costs remarkably less than established competitors globally. According to Artificial Analysis research cited by Reuters, the Chinese startup’s model operates at just 3 cents per benchmark test. Therefore, DeepSeek V4-Flash represents the least expensive option among widely known AI models currently tested. Meanwhile, Anthropic’s Claude Fable 5 costs $3.15 per test—over 100 times more expensive.
DeepSeek Returns to Its Low-Cost Strategy
DeepSeek released V4-Flash officially on Friday pursuing its signature approach. The company seeks to regain momentum through affordable alternatives. Additionally, sources told Reuters the company is preparing for a potential initial public offering. DeepSeek attracted global attention earlier when its R1 model became wildly popular. Furthermore, that release triggered technology stock selloffs internationally.
V4-Flash Pricing
The pricing structure proves remarkably competitive. DeepSeek charges $0.14 per million input tokens and $0.28 per million output tokens. However, Artificial Analysis measured actual operational costs differently. The research firm estimated DeepSeek V4-Flash at 3 cents average per benchmark test.
Comparisons with competitors highlight the advantage clearly:
- Moonshot AI Kimi K3: 86 cents per test
- OpenAI GPT-5.6 Sol: $1.86 per test
- Anthropic Claude Fable 5: $3.15 per test
Test-based comparisons offer more realistic value measurement than listed token prices. Low-priced tokens don’t guarantee efficiency if models require significantly more processing steps. Therefore, total operational cost matters more than unit pricing.
DeepSeek Faces Growing Competition
Competition within China’s AI market intensified dramatically. DeepSeek once dominated attention surrounding Chinese AI development. However, rival startups quickly challenged its position. Moonshot, MiniMax, and Z.AI emerged as serious competitors. Meanwhile, technology giants ByteDance and Alibaba entered the market forcefully.
Chinese companies compete directly with US firms for global adoption. They target businesses seeking affordable AI deployment at scale. Therefore, cost leadership becomes the primary competitive advantage strategically.
V4-Flash Benchmark Performance
Artificial Analysis scored DeepSeek V4-Flash at 50 out of 100. The Intelligence Index combines results from nine benchmarks covering coding, reasoning and workplace assignments. Google’s Gemini 3.6 Flash received the same score equally. Meta’s Muse Spark 1.1 and Zhipu’s GLM-5.2 scored marginally higher.
However, premium models significantly outperform:
- Moonshot Kimi K3: 57 points
- Anthropic Claude Opus 5: 59+ points
- Anthropic Claude Fable 5: 59+ points
- OpenAI GPT-5.6: 59+ points
The performance gap demonstrates that cost leadership involves trade-offs necessarily. DeepSeek V4-Flash prioritizes affordability over raw intelligence metrics.
DeepSeek Preparing V4-Pro
The company is developing a more powerful model called V4-Pro. However, no official release date has been announced yet. Meanwhile, Alibaba unveiled Qwen3.8-Max on Monday. Alibaba described it as its largest and most capable AI model to date.
Finally, DeepSeek V4-Flash pricing strategy forces the entire industry to reconsider value propositions fundamentally.
