Models / DeepSeek pricing
DeepSeek API pricing
7 DeepSeek models from DeepSeek on one key, per million tokens. Input 40% to 90% under DeepSeek's list price, output 51% to 95%. Lowest price: DeepSeek V4 Flash 0423, $0.0279 in and $0.0558 out.
Prices · per million tokens, USD, next to DeepSeek's list price
| Model | Input | Cached input | Output | Under list |
|---|---|---|---|---|
| DeepSeek V4 Pro 0813deepseek-v4-pro-0813 | $0.75list $1.32 | $0.0625list $0.044 | $1.188list $3.96 | 43% in · 70% out |
| DeepSeek V4.1 Flashdeepseek-v4.1-flash | $0.06list $0.30 | $0.0063list $0.006 | $0.18list $1.20 | 80% in · 85% out |
| DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017 | $0.045list $0.30 | $0.0018list $0.006 | $0.18list $1.20 | 85% in · 85% out |
| DeepSeek V4 Flash 0423deepseek-v4-flash | $0.0279list $0.30 | $0.0056list $0.006 | $0.0558list $1.20 | 90% in · 95% out |
| DeepSeek V4 Flash 0731deepseek-v4-flash-0731 | $0.084list $0.14 | $0.0033list $0.0028 | $0.09list $0.28 | 40% in · 67% out |
| DeepSeek V4 Flash Vision Expdeepseek-v4-flash-vision-exp | $0.132list $0.30 | $0.0042list $0.006 | $0.396list $1.20 | 56% in · 67% out |
| DeepSeek V4 Pro 0423deepseek-v4-pro | $0.792list $1.32 | $0.0338list $0.044 | $1.92list $3.96 | 40% in · 51% out |
List price: DeepSeek's published rate, read October 2026. Live from GET https://api.uplinkswitch.com/v1/catalog.
Peak and off-peak, next to one price · per million tokens, input / output
DeepSeek's own API has charged two rates since 2026-08-16: peak, 01:00 to 04:00 and 06:00 to 10:00 UTC on weekdays outside Chinese public holidays, and half that the rest of the time. The rates in force date from 2026-09-10 (DeepSeek pricing, read 2026-10-19). A price here is the same at any hour.
| Model | DeepSeek peak | DeepSeek off-peak | Here, any hour |
|---|---|---|---|
| DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017 | $0.30 / $1.20 | $0.15 / $0.60 | $0.045 / $0.18 |
| DeepSeek V4.1 Flashdeepseek-v4.1-flash | $0.30 / $1.20 | $0.15 / $0.60 | $0.06 / $0.18 |
| DeepSeek V4 Flash 0423deepseek-v4-flash | $0.30 / $1.20 | $0.15 / $0.60 | $0.0279 / $0.0558 |
| DeepSeek V4 Flash Vision Expdeepseek-v4-flash-vision-exp | $0.30 / $1.20 | $0.15 / $0.60 | $0.132 / $0.396 |
| DeepSeek V4 Pro 0813deepseek-v4-pro-0813 | $1.32 / $3.96 | $0.66 / $1.98 | $0.75 / $1.188 |
| DeepSeek V4 Pro 0423deepseek-v4-pro | $1.32 / $3.96 | $0.66 / $1.98 | $0.792 / $1.92 |
On DeepSeek V4.1 Flash 1017, 1,000 requests of 2,000 input and 500 output tokens cost $1.20 at DeepSeek's peak rate, $0.60 off-peak and $0.18 here.
On DeepSeek V4 Pro 0813, 1,000 requests of 2,000 input and 500 output tokens cost $4.62 at DeepSeek's peak rate, $2.31 off-peak and $2.09 here.
What DeepSeek’s API names run · and the id for each here
| DeepSeek API name | Model it runs | Id here |
|---|---|---|
deepseek-flash | DeepSeek V4.1 Flash 1017 | DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017 |
deepseek-v4-pro | DeepSeek V4 Pro 0813 | DeepSeek V4 Pro 0813deepseek-v4-pro-0813 |
deepseek-v4-flash, deepseek-v4-flash-vision-exp | Still accepted; DeepSeek serves them with V4.1 Flash 1017 | Each is its own id here |
| V4 Flash 0731 | Not a model name on DeepSeek’s API | DeepSeek V4 Flash 0731deepseek-v4-flash-0731 |
DeepSeek's API takes two model names today, and what they run moves with each release. The dated ids here (1017, 0813, 0731) each stay on their build, so code you tested keeps calling it until you change the id.
Call them · same base URL and key for every model
Python · OpenAI SDK
from openai import OpenAI
client = OpenAI(
base_url="https://api.uplinkswitch.com/v1",
api_key="sk-uplink-live-...",
)
completion = client.chat.completions.create(
model="deepseek-v4-pro-0813",
messages=[{"role": "user", "content": "Summarize this in one line."}],
)
print(completion.choices[0].message.content)TypeScript · OpenAI SDK
import OpenAI from 'openai';
const client = new OpenAI({
baseURL: 'https://api.uplinkswitch.com/v1',
apiKey: 'sk-uplink-live-...',
});
const completion = await client.chat.completions.create({
model: 'deepseek-v4-pro-0813',
messages: [{ role: 'user', content: 'Summarize this in one line.' }],
});
console.log(completion.choices[0].message.content);Questions
- Which DeepSeek model costs the least?
- DeepSeek V4 Flash 0423 (deepseek-v4-flash): $0.0279 per million input tokens and $0.0558 per million output tokens.
- How far under DeepSeek's list price is it?
- Input runs 40% to 90% under DeepSeek's list price, output 51% to 95%, depending on the model. Each row above shows both prices.
- Is there a subscription?
- No. Prepaid credit from $10, with no subscription or monthly fee. Each top-up locks your prices for 30 days.
- How is cached input billed?
- At the cached rate in the table. A model with no cached rate bills cached tokens at its input rate.
- Are there rate limits?
- Every key carries a request-per-minute limit and a concurrency limit.
- Is the DeepSeek API free?
- No. Creating a key is free, and every token bills from a balance you top up first. Here DeepSeek models bill from prepaid credit from $10; the lowest price is DeepSeek V4 Flash 0423 at $0.0279 input and $0.0558 output per million tokens.
- Is the DeepSeek API cheaper than ChatGPT’s?
- Yes, for the same work. 1,000 requests of 2,000 input and 500 output tokens cost $0.18 on DeepSeek V4.1 Flash 1017 and $2.70 on GPT-6.1 Sol, both at our live prices.
- What is the new pricing for DeepSeek V4 Pro?
- DeepSeek lists V4 Pro at $1.32 input and $3.96 output per million tokens at peak and half that off-peak (DeepSeek pricing, read 2026-10-19). Here DeepSeek V4 Pro 0813 is $0.75 and $1.188 at any hour.
- Does DeepSeek limit API requests?
- DeepSeek caps each account at 2,500 requests in flight on deepseek-flash and 500 on deepseek-v4-pro, and answers HTTP 429 past that (DeepSeek docs, read 2026-10-19). Here each key carries its own requests-per-minute and concurrency limit.