According to TechNode, China's DeepSeek is moving its V4 model from preview to official release in mid-July, bundled with an unusual pricing change: peak-hour API rates that cost roughly double the off-peak price, with peak windows running about 9am-noon and 2-6pm Beijing time. The lineup pairs V4 Pro, a mixture-of-experts model with around 1.6 trillion total and 49 billion active parameters, with the lighter V4 Flash at 284 billion total and around 13 billion active, both carrying a 1-million-token context window and gains in agentic tasks, math and code. Time-of-day pricing is rare among frontier labs and could nudge rivals to rethink how they charge for heavy usage.
NewsAI-assisted
DeepSeek V4 Goes Official With Surge Pricing
China's DeepSeek adds peak-hour API rates at 2x off-peak as it ends its V4 preview.