DeepSeek Launches V4 Flash and V4 Pro — The Biggest Open-Weight AI Models Available Today

[encrypted]

Chinese AI lab DeepSeek has released two preview versions of its newest large language model, DeepSeek V4, marking a significant update to last year's V3.2. Both V4 Flash and V4 Pro are mixture-of-experts models featuring 1 million token context windows, large enough to process entire codebases or lengthy documents in a single prompt. The Pro model is the standout, packing 1.6 trillion total parameters with 49 billion active — making it the largest open-weight model currently available, surpassing Moonshot AI's Kimi K 2.6 at 1.1 trillion and more than doubling DeepSeek's own V3.2 at 671 billion.

DeepSeek claims both models have nearly closed the gap with today's leading closed-source models on reasoning benchmarks, with V4-Pro-Max reportedly outperforming OpenAI's GPT-5.2 and Gemini 3.0 Pro on certain tasks. Coding performance on both models is described as comparable to GPT-5.4. However, the models do trail frontier models in knowledge tests, with DeepSeek acknowledging a developmental gap of approximately three to six months behind state-of-the-art closed models like GPT-5.4 and Gemini 3.1 Pro.

One notable limitation is that both V4 models support text only, unlike many closed-source competitors that handle audio, video, and image inputs. DeepSeek says architectural improvements over V3.2 make both models more efficient and performant, though developers looking for multimodal capabilities will need to look elsewhere for now.

[bonk]

ai models now competing on who can read the longest document without crying