We charge exactly 50% of DeepSeek’s published rate on input and output — $0.150 against their $0.300 on DeepSeek V4.1 Flash, and $0.090 off-peak. Call it through an OpenAI-compatible endpoint: create a key, change one base URL, and keep every line of the client code you have already written. No waitlist, no subscription, no minimum spend.
Per million tokens, billed per token. Cache reads are itemised separately and billed at the model vendor's published cache rate.
| Model | Input / 1M | Output / 1M | Cache read / 1M | Context | |
|---|---|---|---|---|---|
| DeepSeek V4.1 Flash deepseek/deepseek-v4.1-flash50% off | $0.300$0.150 $0.090 off-peak | $1.200$0.600 $0.360 off-peak | $0.006 $0.003 off-peak | 1049K | Use this model |
| DeepSeek V4 Pro deepseek/deepseek-v4-pro50% off | $1.320$0.660 | $3.960$1.980 | $0.044 | 1049K | Use this model |
| DeepSeek V4 Flash deepseek/deepseek-v4-flash50% off | $0.440$0.220 | $1.320$0.660 | $0.014 | 1049K | Use this model |
Struck-through figures are the model vendor’s own published rate; the price beside them is what we charge. Billed per token, no minimum. Cache reads are the exception and carry no discount — they already cost between a fifteenth and a twenty-fifth of the input rate, and we pass the supplier’s cache rate through rather than flattening it into one headline number.
Off-peak pricing. DeepSeek V4.1 Flash is billed at two rates. Peak hours are Mon-Fri 09:00-12:00 and 14:00-18:00 (UTC+8); everything else is off-peak, including weekends. A request is billed at the rate in force when it is made, and every request in your logs records which one it was.
The catalogue
One endpoint and one key for all of them. Prices come from the live catalogue.
deepseek/deepseek-v4.1-flashFast multimodal model. Reads text, images and video; cheapest per token we sell.
Peak is Mon-Fri 09:00-12:00 and 14:00-18:00 (UTC+8); every other hour is off-peak, priced by when the request arrives.
DeepSeek V4.1 Flash API details →deepseek/deepseek-v4-proFrontier reasoning model. Best for agents, long-context analysis and code generation.
deepseek/deepseek-v4-flashHigh-efficiency workhorse. Classification, extraction, chat and high-volume pipelines.
All models, including GLM · running Claude Code on these models
Step by step
No waitlist, no sales call and no regional card check. The steps are the whole process.
Email and a password, or Google. Verifying your email grants free trial credit, so the first call costs nothing.
From the dashboard, one click. Scope it to a single model if you want to, and revoke it the same way.
Set base_url to https://api.tokenify.dev/v1 and use the key as your API key. Model ids are the vendor/model strings shown above.
Card or USDT/USDC. Billing is per token from prepaid credit — no subscription, no minimum, and the credit does not expire.
Two lines
Nothing else in your code changes — streaming, tool calling and JSON mode all pass through untouched.
from openai import OpenAI
client = OpenAI(
base_url="https://api.tokenify.dev/v1", # the only line that changes
api_key=os.environ["TOKENIFY_API_KEY"],
)
response = client.chat.completions.create(
model="deepseek/deepseek-v4-pro",
messages=[{"role": "user", "content": "Hello"}],
)Create a Tokenify account with email or Google and generate a key from the dashboard. There is no waitlist and no regional card requirement. The key works with the standard OpenAI SDK by changing the base URL.
DeepSeek V4 Flash costs $0.220 per million input tokens against DeepSeek's own list price of $0.440. V4 Pro costs $0.660 against a list price of $1.320. That is 50% off on input and output, on every model, with no volume tier to reach. Billing is per token with no subscription or minimum, and cached prompt tokens are billed at a fraction of fresh ones.
No. Tokenify is an independent reseller. We buy capacity wholesale and route each request to the cheapest healthy source, which is where the discount comes from. DeepSeek is a trademark of its owner and this page is not affiliated with or endorsed by them.
No. Verify your email and you get free trial credit. Add a card only when you want more.
A circuit breaker takes the failing source out of rotation within seconds and the request is retried before any bytes reach your client, so where a model has another source you see a slightly slower response instead of an error. Where every source has been tried and none answered, you get a 502 — or a 429 if that is what the last attempt said — naming the model, and the request is not charged.
Yes — the same key and models are served at /v1/messages in the Anthropic format.
DeepSeek prices by the clock and the resellers do not all price the same way. DeepSeek API pricing, compared puts the official rates, the other sellers and ours in one table, with the date each figure was read, and costs three months of real traffic at each.
Last updated 2026-10-01. Prices come from the live catalogue at the time this page was built.