DeepSeek's pricing page lists two chat models on its API, deepseek-flash and deepseek-v4-pro, each with a peak and an off-peak rate, plus standalone Search and Fetch APIs billed per request. All figures below were checked against DeepSeek's API documentation as of 10/22, Beijing time. Prices are in US dollars per million tokens.
Prices per model
The model name deepseek-flash is now served by DeepSeek-V4.1-Flash-1017, the update covered in our look at DeepSeek-V4.1-Flash-1017. Off-peak, it costs $0.003 per million input tokens on a cache hit, $0.15 on a cache miss and $0.60 for output. At peak, those rates are $0.006, $0.30 and $1.20.
deepseek-v4-pro, listed as DeepSeek-V4-Pro-0813, costs $0.022 for cache-hit input, $0.66 for cache-miss input and $1.98 for output off-peak, and $0.044, $1.32 and $3.96 at peak. DeepSeek's September announcement says deepseek-v4-pro requests are routed to V4.1-Flash at Flash rates until V4.1-Pro launches, while its pricing page still lists separate V4-Pro-0813 rates; DeepSeek hasn't said which applies now.
| Model | Cache-hit input | Cache-miss input | Output |
|---|---|---|---|
| deepseek-flash, off-peak | $0.003 | $0.15 | $0.60 |
| deepseek-flash, peak | $0.006 | $0.30 | $1.20 |
| deepseek-v4-pro, off-peak | $0.022 | $0.66 | $1.98 |
| deepseek-v4-pro, peak | $0.044 | $1.32 | $3.96 |
Both models list a 1M-token context window and up to 384K output tokens. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but DeepSeek says those requests are served by DeepSeek-V4.1-Flash-1017 and billed at the Flash price. Images sent to the vision model were billed as up to 384 tokens each when DeepSeek launched image input in August, and the Files API is free to use.
Peak and off-peak hours
DeepSeek says off-peak rates are half of peak rates. Peak hours are 09:00 to 12:00 and 14:00 to 18:00 Beijing time, Monday through Friday, excluding Chinese public holidays. Every other hour is off-peak, including weekends and Chinese public holidays in full.
Search and Fetch
Search and Fetch are standalone APIs, each priced at $0.75 per 1,000 successful requests; only successful requests are charged. When the DeepSeek API is used through DeepSeek Harness, Claude Code or Codex, server-side tool calls already use these APIs. Those calls are free for now, and per-request billing starts at 00:00 Beijing time on November 1, 2026. How the two APIs work is covered in our guide to DeepSeek's Search and Fetch APIs.
Free credits and billing
The cost of a request is the number of tokens times the price. Charges come out of a topped-up balance or a granted balance, and the granted balance is used first when an account has both. DeepSeek does not publish whether new accounts receive a granted balance, how much it is, or when it expires. DeepSeek also says it may adjust prices and recommends checking the pricing page regularly.
Rate limits
DeepSeek limits concurrency rather than requests per minute. Each account can run 2,500 concurrent requests on deepseek-flash and 500 on deepseek-v4-pro, counted across all API keys. A request over the limit gets an HTTP 429 error. Higher concurrency can be requested at no additional cost. DeepSeek does not publish a token-per-minute or daily cap.
Recent price changes
- 00:00 Beijing time, August 17, 2026: with the DeepSeek-V4-Pro launch, DeepSeek updated API prices and introduced peak and off-peak rates, with off-peak 50% below peak.
- 12:00 Beijing time, September 10, 2026: new, lower prices took effect with DeepSeek-V4.1-Flash, which DeepSeek said lets it serve more users at a lower cost.
- 12:00 Beijing time, September 14, 2026: DeepSeek said all deepseek-v4-pro requests would route to V4.1-Flash at V4.1-Flash rates until V4.1-Pro launches.
- October 17, 2026: Search and Fetch opened to everyone at $0.75 per 1,000 calls per endpoint.
DeepSeek's current pages don't list the rates that applied before each change.
How it compares with Claude and OpenAI
Anthropic's cheapest current model, Claude Haiku 5.5, costs $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, and Claude Sonnet 5.5 costs $2 and $10, according to our Claude API pricing guide. OpenAI's gpt-6.1-sol costs $2 for input, $0.10 for cached input and $10 for output, and gpt-6-luna costs $0.10, $0.01 and $0.50, as listed in our GPT-6 Sol and Luna comparison.
At peak, deepseek-flash's $0.30 cache-miss input and $1.20 output are above Haiku 5.5 and gpt-6-luna but well below Sonnet 5.5 and gpt-6.1-sol. On cache hits, deepseek-flash stays cheaper even at peak: $0.006 against $0.01 for Haiku 5.5 cache reads and gpt-6-luna cached input. Token counts differ between model families, so equal prices per token don't mean equal cost per task.