DeepSeek has released an upgraded version of its flagship large language model, dubbed V4 Pro, according to a report from Decrypt. The Chinese AI developer has built its reputation on delivering competitive performance at a fraction of the cost charged by Western rivals.
According to the report, V4 Pro closes much of the performance gap with Anthropic's Claude Fable model. The two systems are said to differ by only around 5% on the benchmarks referenced in the report. That is a notably small margin given the price difference between the two products.
Decrypt reports that Claude Fable costs roughly 4,500% more to use than DeepSeek's V4 Pro. In practical terms, that means Anthropic's model is priced at about 45 times the rate of DeepSeek's offering for comparable output. The disparity underscores a persistent theme in the AI sector: incremental gains in model quality often come with disproportionate increases in cost.
DeepSeek first drew global attention in early 2025 when it demonstrated that highly capable AI models could be trained and served far more cheaply than assumed. That earlier disruption rattled valuations across the AI supply chain, from chipmakers to cloud providers, as investors reassessed how much computing power was truly necessary to match top-tier performance.
The latest upgrade, as described by Decrypt, suggests DeepSeek is continuing that strategy rather than chasing outright benchmark leadership. Instead of matching or beating Claude Fable outright, V4 Pro appears designed to offer a cost-efficient alternative that comes close enough on quality to make the price gap the more relevant factor for many users.
Pricing structures for AI models vary by provider and can include differences in context window size, latency, and enterprise support, none of which were detailed in the available reporting. Readers should note that the comparison figures cited here come from a single published account and have not yet been independently verified across other outlets.
Market Impact
If the reported price and performance gap holds up, it could intensify pressure on Western AI labs to justify premium pricing for incremental model improvements. Enterprises and developers weighing AI infrastructure costs often factor in exactly this kind of price-to-performance ratio when choosing providers.
The broader crypto and tech investment community has previously reacted sharply to DeepSeek news, given the company's history of triggering swings in AI-linked equities and tokens tied to compute demand. Any renewed narrative around cheaper, competitive AI models from China could again draw attention to AI-adjacent crypto assets, though no specific market reaction has been confirmed at this time.
The reported gap between DeepSeek's V4 Pro and Anthropic's Claude Fable highlights an ongoing cost debate in the AI industry. Further reporting will likely clarify the benchmarks and pricing details behind these claims.
Frequently Asked Questions
What is DeepSeek V4 Pro?
It is an upgraded large language model released by the Chinese AI company DeepSeek, according to a Decrypt report.
How does V4 Pro compare to Claude Fable?
Decrypt reports that Claude Fable, an Anthropic model, performs only about 5% better than V4 Pro while costing roughly 4,500% more to use.
Why does the price difference matter?
A large cost gap for a small performance improvement raises questions about whether premium AI pricing is justified, which matters for businesses choosing between providers.
Has this comparison been independently verified?
The figures come from a report by Decrypt, and readers should treat specific benchmark and pricing details as pending further confirmation.