China’sDeepSeekhas released its long-awaited newartificial intelligence modelV4, saying it offers world-beating capabilities and that a preview version is now available to use.
The versionDeepSeek-V4 is reportedly better optimised forChina’s domestic chips.
It "features an ultra-long context of one million words, achieving leadership in both domestic and open-source fields across agent capabilities, world knowledge, and reasoning performance",DeepSeeksaid in a statement.
The new model is available in two versions,DeepSeek-V4-Pro andDeepSeek-V4-Flash, according to the company, which says the latter is a "more efficient and economical choice".
"In world knowledge benchmarks, DeepSeek-V4-Pro significantly leads other open-source models and is only slightly outperformed by the top-tier closed-source model, (Google's) Gemini-Pro-3.1," the Hangzhou-basedAIcompany said.
DeepSeek-V4-Pro also comes with a “maximum reasoning effort mode”, which the AI startup claims “significantly advances the knowledge capabilities of open-source models, firmly establishing itself as the best open-source model available today”.
The latest update comes after the company sparked a trillion-dollar stock market selloff last year following the release of its R1 model, which rivalled the performance of AI systems like ChatGPT while built for only a fraction of the cost.
The 2025 release sent shockwaves throughout the tech industry, with Nvidia experiencing its biggest one-day loss in market value in history, losing more than $500bn, while other tech giants like Oracle, Amazon and Microsoft also experienced significant drops in share price.
This was because the DeepSeek model was free-to-use, open-source, and developed at only a fraction of the cost, and the first time a Chinese competitor could rival the most advanced AI models from US giants.
Advertisement
DeepSeek’s latest release comes amid growing semiconductor export restrictions from the US to China, especially high-end graphical processing units (GPUs), which are key for building AI models, forcing China to rely on its own homegrown GPU manufacturers.
The company did not unveil what chip system it used to train the V4 models, but said its software components are designed to work with both Nvidia and Huawei chips.
So far, the AI firm has only released basic details of the new version, including that it has the ability to process a maximum output of 384,000 tokens.
Tokens are the basic unit of data that AI models process, and can be words or characters. Typically, a token can be about 4 characters, and the faster a model processes tokens, the faster it can learn and respond.
The Chinese AI firm says the new version achieves a “dramatic leap in computational efficiency” with its ability to process and understand the context of up to 1 million tokens.
In comparison, the previous version, V3, was able to understand the context of up to 128,000 tokens.
The new upgrade enables multi-document reasoning, with the AI model now capable of understanding the context of entire books and full code databases.
“This breakthrough enables efficient support for a context length of one million tokens, ushering in a new era of million-length contexts for next-generation large language models,” it said.
In terms of understanding the context of long strings of text, DeepSeek-V4-Pro outperforms Google’s Gemini-3.1-Pro, the company said, adding, however, that it remains behind Anthropic’s Claude Opus 4.6 AI model.
DeepSeek said it hopes to further enhance the model’s intelligence, robustness, and practical usability across a broad range of scenarios and tasks.