DeepSeek-V4 Pro now available on Together AI — 512K context, long-context reasoning

AI Impact Summary

Together AI has launched DeepSeek-V4 Pro, a 1.6T-parameter MoE model with a 512K context window, designed for long-context reasoning workloads like code agents and document intelligence. This release leverages Together AI's Serverless Inference and Monthly Reserved infrastructure, offering a path from experimentation to production, particularly focusing on cost optimization through cached input pricing. The model’s hybrid attention architecture and three reasoning modes (Non-Think, Think High, Think Max) provide flexibility for diverse tasks, while the benchmark results demonstrate strong performance across coding, reasoning, and agentic tasks.