Paid links may earn Compute Screener a referral fee. They never affect prices, weights or the order.
About DeepSeek
The DeepSeek token price index is $0.28 per 1M tokens today, down 2.2% since yesterday, down 1.3% over the past week and down 7.8% over the past month. Since Aug 17, 2026 it has ranged from $0.28 to $0.40. It is the usage-weighted average of the posted prices of DeepSeek's 2 models, blending 3 input tokens with 1 output token; DeepSeek V4.1 Flash 1017 has the largest share of usage, at 97.6%. It is calculated once a day, for each UTC day, shortly after the day ends.
DeepSeek is an AI lab in Hangzhou, founded on July 17, 2023 and owned and funded by High-Flyer, a Chinese hedge fund; its founder, Liang Wenfeng, is chief executive. It releases its models as open weights under the MIT license, and it led all model makers on OpenRouter with 24.4% of requests in the week of October 12, 2026. Its prices change with the time of day: since August 16, 2026 they are halved outside peak hours.
Models and list prices
| Model | Input | Output | 3:1 blend |
|---|
| DeepSeek V4.1 Flash, off-peak | $0.15 | $0.60 | $0.2625 |
| DeepSeek V4.1 Flash, peak | $0.30 | $1.20 | $0.525 |
| DeepSeek V4 Pro, off-peak | $0.66 | $1.98 | $0.99 |
| DeepSeek V4 Pro, peak | $1.32 | $3.96 | $1.98 |
USD per 1M tokens on October 24, 2026, from DeepSeek's pricing page. Peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday; "All other hours are off-peak, including weekends and Chinese public holidays in full." Cached input costs 2% of the regular input price on Flash and about 3.3% on V4 Pro, and both models have a 1M-token context window. Calls made through DeepSeek Harness, Claude Code or Codex are free until 16:00 UTC on October 31, 2026.
What this index measures
Each day's value averages DeepSeek's off-peak prices for the models people use, weighted by each model's tokens in OpenRouter's daily rankings and blended at 3 input tokens to 1 output token. DeepSeek's own endpoint on OpenRouter listed the off-peak rate when checked; work run in peak hours costs twice as much. On October 23, 2026, V4.1 Flash carried 97.6% of the tokens the index counts for the models DeepSeek itself serves, so the index sat close to Flash's off-peak blend.
Other hosts sell the same open weights. On October 24, 2026, 34 endpoints listed V4.1 Flash on OpenRouter, from $0.02 to $0.41 per million input tokens, and several large hosts, including Together and Fireworks, listed $0.30/$1.20, DeepSeek's peak price. The index uses only DeepSeek's own price.
What moves DeepSeek's index
- DeepSeek's price cuts. DeepSeek released V4.1 Flash on September 10, 2026 with lower prices: "We're passing the savings on to you." The index fell from $0.3986 on September 9 to $0.3096 on September 12.
- Flash versus Pro. V4 Pro's blended price is almost four times Flash's, so the index rises when Pro's share of usage rises. Pro carried 2.4% of the tokens the index counted for DeepSeek on October 23, 2026.
- Version switches. DeepSeek serves only its latest versions; older model names are "served by the DeepSeek-V4.1-Flash-1017 model and billed at the Flash price", so a new version's price applies to the old names too.
- Peak hours. The index uses off-peak rates. Peak hours are 35 of the week's 168, and in them every price is twice the off-peak rate.
Price history
- August 2, 2024: context caching launched; cache hits cost $0.014 per million tokens against $0.14 for other input.
- February 8, 2025: DeepSeek V3, or deepseek-chat, rose from $0.14/$0.28 to $0.27/$1.10.
- September 5, 2025: new prices for V3.1 took effect and DeepSeek's earlier off-peak discounts ended.
- September 29, 2025: "DeepSeek API prices drop 50%+, effective immediately", with V3.2-Exp.
- April 24, 2026: V4 Preview released, with a 1M-token context by default.
- August 16, 2026: new prices with peak and off-peak rates, off-peak 50% lower, from 16:00 UTC.
- September 10, 2026: V4.1 Flash, with lower prices from 04:00 UTC.
- October 17, 2026: V4.1-Flash-1017 replaced it; "The API calling method and pricing remain unchanged".
This index's history for DeepSeek starts on August 17, 2026, the first full day of its current price structure.
US government actions
On October 20, 2026 the Department of Homeland Security published an order excluding DeepSeek's products from federal civilian agency procurement and removing "all DeepSeek and DeepSeek-affiliate products, and third-party products and services developed using DeepSeek implementations, from civilian agency information technology". The 2026 National Defense Authorization Act also requires AI developed by DeepSeek to be removed from devices operated by the Defense Department, the intelligence community or their contractors, according to Wikipedia, citing the Congressional Research Service.
FAQ
How much does the DeepSeek API cost per million tokens? DeepSeek V4.1 Flash costs $0.15 input and $0.60 output off-peak and twice that at peak; V4 Pro costs $0.66 and $1.98 off-peak, as in the table above.
When are DeepSeek's peak hours? 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays, which is 9:00 to 12:00 and 14:00 to 18:00 in Beijing. Weekends and Chinese public holidays are off-peak all day.
Why does the index use off-peak prices? They are the rates DeepSeek's own endpoint listed on OpenRouter when checked, and off-peak covers 133 of the week's 168 hours.
How often is it updated? Once a day: each UTC day's value is calculated shortly after that day ends, from the day's OpenRouter volumes and the prices posted then. The site checks for new data every 2 to 3 hours.
Sources
Text reviewed Oct 24, 2026; the figures in it are as of that date. The figures at the top of this page update once a day. Not investment advice.
