According to CoinMeta, DeepSeek has officially launched a new visual model called deepseek-v4-flash-vision-exp. The official API documentation lists it alongside V4 Flash and V4 Pro. Developers can pass in images through DeepSeek API. This model supports up to 1 million token contexts and can output a maximum of 384,000 token. It also supports JSON output, tool calls, response to API, and Anthropic API. The pricing is aligned with V4 Flash; for every million token inputs, the cost during peak hours is 3 yuan, 1.5 yuan during off-peak hours, and 0.1 yuan and 0.05 yuan respectively when the cache is hit. For every million token outputs, the cost during peak hours is 9 yuan, and 4.5 yuan during off-peak hours, which is only one-third of the price of V4 Pro. Images are billed based on their size converted into token. The peak pricing periods are from 9:00–12:00 and 14:00–18:00 Beijing time, while the price is halved during other times.