DeepSeek’s New AI Model Reportedly Costs Far Less to Run Than Leading Rivals

Chinese AI startup DeepSeek has once again upended the global AI market with the release of its V4-Flash model, which costs just $0.14 per million input tokens and $0.28 per million output tokens according to research firm Artificial Analysis.

That pricing makes V4-Flash the least expensive well-known AI model to run, with an average benchmark test cost of approximately $0.03 — about 105 times cheaper than Anthropic's Claude Fable 5, which costs $3.15 per test.

The comparison accounts not just for headline token prices, but also the amount of data a model must process and generate to complete a task, providing a more realistic measure of value.

A model with low headline price can still prove expensive if it requires significantly more steps to produce an answer. V4-Flash, however, consistently delivers tasks at a fraction of the cost of its rivals.

"DeepSeek's V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens. Artificial Analysis estimated V4-Flash's average cost at 3 cents per test, compared with 86 cents for Kimi K3 from Chinese rival Moonshot AI, $1.86 for OpenAI's GPT-5.6 Sol and $3.15 for Claude Fable 5." – Reuters reporting on the DeepSeek pricing comparison

While DeepSeek's R1 model triggered a historic selloff in global technology stocks in early 2025, the company lost momentum as domestic rivals like Moonshot, MiniMax, Z.ai, ByteDance, and Alibaba flooded the market with competitive offerings.

V4-Flash represents DeepSeek's attempt to regain its edge by doubling down on its core strategy: ultra-low-cost AI alternatives. On Artificial Analysis's Intelligence Index, which combines results from nine benchmarks spanning coding, reasoning, and workplace-style assignments, V4-Flash scored 50 out of 100.

That puts it on par with Google's Gemini 3.6 Flash, one point behind Meta's Muse Spark 1.1 and Z.ai's GLM-5.2, but still lagging behind Moonshot's Kimi K3 (57) and the top-tier models from Anthropic and OpenAI, which scored nine or more points higher.

Despite the performance gap, the cost advantage is staggering. For the same coding output that costs $25 on Anthropic's Claude Opus 4.8, DeepSeek charges approximately $0.28 — a difference of roughly 99 percent.

This pricing pressure arrives at a critical moment, as agentic AI workloads that require repeated model calls are exploding. For enterprises building AI agents that must execute complex, multi-step tasks at scale, the total cost of inference can quickly become prohibitive.

DeepSeek's V4-Flash offers a compelling alternative for cost-sensitive deployments where near-frontier performance is sufficient.

As AI cost efficiency becomes a strategic differentiator, DeepSeek's V4-Flash is already reshaping how businesses evaluate AI investments. With DeepSeek reportedly preparing for a potential IPO and a more powerful V4-Pro version on the horizon, the startup is positioning itself to reclaim its role as a disrupter in the global AI landscape. For now, the message is clear: in the race for AI dominance, cost is becoming as critical as capability.