Chinese AI startup DeepSeek has dramatically cut the price of its flagship V4-Pro model by 75%, just one month after releasing the V4 generation of large language models. This aggressive pricing move is reshaping the competitive landscape of enterprise AI, putting pressure on Western providers like OpenAI, Anthropic, and Google to justify their premium pricing models.

The price cut and its significance

DeepSeek announced that the V4-Pro API pricing would be reduced permanently—not as a promotional discount but as a reflection of underlying efficiency gains. Previously, usage costs ranged from $0.0145 per million tokens for cache hits to $3.48 per million output tokens. After the adjustment, prices start at $0.003625 per million tokens for cache hits and go up to $0.87 per million tokens for output. The company plans to officially set the API pricing at one-quarter of the original rates once the promotional period ends on May 31, 2026.

According to Sanchit Vir Gogia, chief analyst and CEO at Greyhound Research, the V4-Pro was engineered to slash the cost of long-context inference. He noted that the model runs at roughly a quarter of the single-token compute and a tenth of the memory footprint of its predecessor at very long contexts. "This is why the price cut is permanent rather than promotional. It is not a discount. It is an efficiency gain being passed through," Gogia explained.

DeepSeek narrows the performance gap

Nearly a year after releasing the R1 reasoning model—which offered competitive performance at lower cost—DeepSeek launched the V4 LLM preview. Like earlier models, V4 is open source, allowing developers to download, modify, and run it locally. The new models are optimized for popular agent tools such as Anthropic’s Claude Code and OpenClaw.

Neil Shah, vice president at Counterpoint Research, said that from a pure capabilities standpoint, DeepSeek V4-Pro has effectively closed the performance gap on critical tasks like complex math and reasoning. "It aggressively leads the market on openness and inference costs. Its specialized reasoning modes and architectural enhancements make it a formidable alternative to Western frontier models," Shah noted. However, he added that the model’s primary limitations are not raw intelligence but rather ecosystem adoption, global support structures, IP provenance, and deep hyperscaler integrations that AWS, Microsoft, and Google natively provide.

Enterprise cost implications

Inference costs remain one of the biggest barriers to scaling AI pilots into organization-wide deployments. DeepSeek’s steep discounts could translate into substantial savings for enterprises, provided they can access the model at scale.

Amit Jaju, senior managing director at Ankura Consulting, pointed out that for most enterprises, the relevant comparison is not DeepSeek’s direct API but the cost of running a local deployment versus using any external inference provider. "If a CIO can host DeepSeek V4-Pro on their own infrastructure, inference costs drop dramatically, and many projects that were previously uneconomical become viable. That includes always-on copilots, bulk document review, code generation, L1 support, and multi-agent workflows," Jaju explained. He warned that if the model is consumed through third-party providers, the effective rate may be higher and the ROI benefit smaller.

The first wave of enterprise AI was full of impressive demonstrations and uncomfortable invoices, added Gogia. He noted that CIOs quickly learned that the cost of AI was never just the model call but included retrieval, orchestration, and more.

Pressure on premium AI pricing models

DeepSeek’s pricing strategy is likely to intensify pressure on major AI vendors whose models often command premium enterprise pricing. This could prompt OpenAI, Anthropic, and Google to offer better packages or shift to value-based pricing.

Shah observed that high-margin, high-consumption token pricing from Anthropic and OpenAI is becoming harder to justify for many enterprise workloads. "The presence of a viable open-weights alternative gives enterprise buyers decent leverage. This will likely prompt premium Western AI labs to gradually shift from basic consumption-based pricing toward more defensible, outcome-oriented monetization models," he said.

As a result, CIOs are expected to adopt multi-model AI strategies, similar to the migration to multi-cloud architectures. Gogia described this as an AI portfolio architecture where premium models are used for high-stakes work, domain models for specialist tasks, smaller models for repeatable execution, and an orchestration layer to route, log, govern, and monitor the entire estate.

Risks of adopting Chinese-origin AI models

Despite the cost advantages, CIOs must proceed cautiously when evaluating Chinese-origin AI models. Key risks include data sovereignty, cross-border exposure, IP leakage, and regulatory defensibility.

Jaju highlighted that the primary risk is data sovereignty. If CIOs rely on external APIs hosted in China, prompts, documents, embeddings, logs, and telemetry can leave the enterprise perimeter and traverse jurisdictions with different legal regimes. IP leakage is another serious concern, as developers may paste source code, product designs, legal drafts, or incident data into model workflows. If the model is external, that data can be stored, used for training, or exposed through logs or plugins.

Regulatory defensibility is a third risk. CIOs need clarity on where data is processed, what is retained, who can access it, what contractual protections exist, whether the model can be self-hosted, and how outputs can be audited. Experts warn that the safest approach is to host DeepSeek locally or in a sovereign cloud under enterprise control, with encryption, access controls, and audit trails.

Industry-wide implications

DeepSeek’s move is part of a broader trend of falling inference costs driven by architectural innovations and intense competition. The company’s open-source strategy and aggressive pricing are forcing established players to rethink their pricing and product strategies. For enterprises, the opportunity to drastically reduce AI costs is real, but it comes with trade-offs in terms of data security, compliance, and integration.

The AI pricing war shows no signs of abating. As more efficient models emerge and competition intensifies, the balance of power between proprietary and open-source AI continues to shift. Enterprises that carefully evaluate both cost and risk will be best positioned to harness the benefits of this rapidly evolving landscape.

Source: InfoWorld News