Handles repository-scale and long-document prompts with a 384K output ceiling.
Back to models
Million-token context Image understanding Off-peak discount
DeepSeek
DeepSeek V4.1 Flash 1017
FeaturedFrontier context at Flash economics44% offInputImageTextOutputText
API formatOpenAI
A high-throughput DeepSeek model with a 1M-token context window, image understanding, and peak/off-peak pricing.
deepseek-v4.1-flashInput price$0.1571$0.28169per million tokens
Output price$0.6285$1.126761per million tokens
Cache read$0.00314$0.005634per million tokens
Context window1.0M tokensMaximum context per request
Max output393.2K
BillingPer token
HealthOperationalSuccess 100% · latency 802ms · throughput 6.1
Model details
Capabilities, benchmarks, and typical use
DeepSeek V4.1 Flash is DeepSeek's high-throughput model, serving requests under the deepseek-v4.1-flash name. Thinking mode is enabled by default and can be switched to non-thinking mode. It supports JSON output, tool calls, the Responses API, the Anthropic-compatible API, conversation prefix continuation, FIM completion in non-thinking mode, and image understanding. The context window is 1M tokens with a maximum output of 384K tokens. DeepSeek bills by time of day: weekday 09:00-12:00 and 14:00-18:00 Beijing time are peak hours with the full rate, and all other periods including weekends and public holidays are charged at half. The original prices shown here are the peak-hour CNY list prices converted at 7.1 CNY per USD: CNY 2.00 per 1M cache-missed input tokens, CNY 0.04 per 1M cache-hit input tokens, and CNY 8.00 per 1M output tokens. Off-peak rates are half of these values.
Accepts image input alongside text for multimodal analysis.
Costs half the peak rate outside weekday business hours.