DeepSeek Unveils V4 Pro 0813 AI Model with Extended Context on OpenRouter

DeepSeek has officially released the V4 Pro 0813, a large-scale mixture-of-experts artificial intelligence model now accessible via the OpenRouter platform. The launch introduces a model designed to handle extensive data processing tasks, featuring a context window of 1,048,576 tokens and a maximum output length of 384,000 tokens.

The availability of the V4 Pro 0813 through OpenRouter provides developers and enterprises with a centralized point of access to DeepSeek’s latest capabilities. OpenRouter, which serves as a distribution channel for various AI models, has listed the new specifications and pricing for the model, making it immediately available for integration into existing workflows.

Pricing for the new model is set at $0.435 per million input tokens and $0.87 per million output tokens. These figures reflect the current listing on OpenRouter and represent the cost structure for users accessing the model through the platform. As with many AI service providers, these rates are subject to change by the provider.

The introduction of the V4 Pro 0813 aligns with ongoing developments in the machine learning sector, where providers are increasingly focusing on expanding context windows to improve performance on complex, long-form tasks. The model’s architecture, described as a large-scale mixture-of-experts, aims to balance computational efficiency with high-level reasoning capabilities.

While the specific technical architecture details beyond the mixture-of-experts classification are not fully detailed in the source material, the release underscores the competitive pressure in the AI market to deliver models with greater token capacity. Claims regarding the model’s long-term performance and reliability should be verified through independent testing, as the current information stems from a product listing rather than an independent review.

The release of the V4 Pro 0813 follows a period of rapid iteration in the AI industry, with competitors such as Google’s Gemini app recently surpassing one billion monthly active users. As the market continues to evolve, the availability of models like DeepSeek’s new offering provides institutions and developers with additional options for scaling their artificial intelligence applications.

For now, the V4 Pro 0813 stands as a new entry in the large language model landscape, offering a specific combination of context length and output capacity at a defined price point. Users interested in utilising the model can access it directly through the OpenRouter platform, where further technical documentation and usage guidelines are likely to be updated.