DeepSeek V4 Preview Models Released, Competitive with Frontier AI at Lower Cost

Chinese AI lab DeepSeek released the first models in its V4 series on April 24, 2026: DeepSeek-V4-Pro and DeepSeek-V4-Flash. Both are 1 million token context Mixture of Experts models under the MIT license. Pro has 1.6 trillion total parameters (49B active), making it the largest open weights model, surpassing Kimi K2.6 (1.1T) and GLM-5.1 (754B). Flash has 284B total parameters (13B active). Pro is 865GB on Hugging Face; Flash is 160GB. Pricing is significantly lower than competitors: Flash costs $0.14/million input tokens and $0.28/million output; Pro costs $1.74/million input and $3.48/million output. DeepSeek-V4-Flash is the cheapest small model, beating OpenAI's GPT-5.4 Nano; Pro is the cheapest large frontier model. Efficiency improvements include 27% of single-token FLOPs and 10% KV cache size for Pro at 1M context compared to V3.2, and 10% FLOPs and 7% KV cache for Flash. Benchmarks show Pro competitive with GPT-5.2 and Gemini-3.0-Pro, but trailing GPT-5.4 and Gemini-3.1-Pro by 3-6 months. The Unsloth team is expected to release quantized versions soon.