DeepSeek offers 75% discount on V4-Pro as it challenges OpenAI and Google

In the fast-moving world of artificial intelligence, bigger is often seen as better. But cheaper? That is where things get interesting.

Chinese AI startup DeepSeek is attempting to shake up the market not just with a powerful new model, but with aggressive pricing that could force competitors to rethink their strategies. With the launch of its latest model, DeepSeek is signalling that the next phase of the AI race may be fought as much on cost as on capability.

DeepSeek new AI model gets 75% discount

To mark the debut of its new V4-Pro model, DeepSeek is offering developers a steep 75 per cent discount until May 5. Alongside this, the company has slashed the cost of input cache hits across its API suite to just one-tenth of previous prices.

The timing appears strategic. As demand for AI tools grows, so too do concerns around rising usage costs, particularly for developers building large-scale applications. By lowering the barrier to entry, DeepSeek is positioning itself as an attractive option for both startups and enterprises looking to scale efficiently.

More from Tech

The company recently unveiled a preview of the V4 model, which has been adapted to run on hardware from Huawei, highlighting a push towards tighter integration with domestic chip ecosystems.

A look at how DeepSeek's pricing stands against other AI models

DeepSeek’s aggressive pricing puts it in direct competition with some of the biggest names in AI. For context, OpenAI’s GPT-5.5 Pro is priced at roughly $5 per million input tokens and $30 per million output tokens.

Similarly, Anthropic’s Claude Opus 4.7 comes in at around $5 per million input tokens and $25 per million output tokens, while Google’s Gemini 3.1 Pro is priced at $2 per million input tokens and $12 per million output tokens.

Subscription-based platforms such as Perplexity AI also add another layer to the competition, offering bundled access to multiple models for a monthly fee starting at $17.

Against this backdrop, DeepSeek’s discounts could significantly alter cost calculations for developers, particularly those working on AI agents or high-volume applications.

DeepSeek V4 Pro AI model

At the core of this push is the V4-Pro model itself, which DeepSeek describes as its most advanced offering yet. With a staggering 1.6 trillion parameters, it is among the largest open-source models currently available.

However, the model uses a technique known as Mixture-of-Experts, meaning only a fraction of these parameters, around 49 billion, are activated for any given task. This allows the system to maintain high performance without proportionally increasing computational costs.

DeepSeek claims that the Pro variant outperforms most open-source rivals in world-knowledge benchmarks, trailing only top-tier closed models like Google’s Gemini-Pro-3.1. It is also designed with AI agents in mind, enabling more complex, multi-step tasks compared to traditional chatbots.

Early developer feedback appears promising. In an internal survey, a majority of users indicated they would consider making V4-Pro their default coding model, with many noting performance close to leading proprietary systems.

As AI tools become more embedded in everyday workflows, DeepSeek’s strategy suggests a shift in priorities. Performance still matters, but affordability may soon prove just as decisive.