DeepSeek-V4-Pro costs LKR 519 per million input tokens and LKR 1,039 per million output tokens on Roar AI, billed in Sri Lankan rupees on one invoice. DeepSeek’s open flagship: a 1.6-trillion-parameter model under the plain MIT licence.
Prices are shown in LKR to make model costs easier to estimate and manage for Sri Lankan teams, with the published US dollar rate beside each.
Example: 1,000 requests with 3,000 input and 1,000 output tokens each cost LKR 2,597. Every rate is published and stays the same from one request to the next. See every model.
Rates as of 20 November 2026. Rupee prices are our published US dollar rates converted at LKR 333 to the dollar, our rate for November 2026 — the same rate your invoice uses.
deepseek-v4-proTry DeepSeek-V4.1-Flash before you pay for DeepSeek V4 Pro. Flash costs less, reads images as well as text, and DeepSeek says it now beats V4-Pro on its own tests. Those scores have not been independently checked, so compare the two on your own work.
V4-Pro is the larger model, at 1.6 trillion parameters. DeepSeek reports 80.6% on SWE-bench Verified and a Codeforces rating of 3206 in its top reasoning mode, which makes it worth testing on hard coding and maths.
| Model | Input | Output | Choose it for |
|---|---|---|---|
| DeepSeek-V4-Pro | LKR 519 | LKR 1,039 | Coding across large repositories. |
| DeepSeek-V4.1-Flash | LKR 64.94 | LKR 260 | A low-cost default for coding, tool use and images. |
| Kimi K3 | LKR 1,234 | LKR 6,169 | Web research, images and video. |
| GLM-5.3 | LKR 424 | LKR 1,333 | Security analysis and command-line agents. |
Prices per 1M tokens.
DeepSeek-V4-Pro is the largest model DeepSeek has released: 1.6 trillion parameters, 49 billion of them at work on each token, with weights open under the MIT licence, the most permissive terms available on a model this size. DeepSeek reports 80.6% on SWE-bench Verified and a Codeforces rating of 3206 in its top reasoning mode.
Much of the engineering went into making a million-token context affordable: DeepSeek says its new attention design cuts the compute per token to about a quarter of V3.2’s at that length. It has three modes (no thinking, think high and think max) and reads text only.
Roar AI provides access to DeepSeek-V4-Pro through the Roar AI gateway, with usage billed in LKR and no foreign card needed. Developers can use DeepSeek-V4-Pro alongside other supported models without maintaining a separate integration for every provider.
Use the OpenAI SDK you already have: point it at Roar AI and set the model name. Read the docs.
from openai import OpenAI
client = OpenAI(base_url="https://api.roar-ai.com/v1", api_key="roar_live_…")
reply = client.chat.completions.create(
model="deepseek-v4-pro",
messages=[{"role": "user", "content": "Summarise this contract in plain English."}],
)Requests to DeepSeek-V4-Pro are processed in Singapore, outside Sri Lanka.
Roar AI’s locally hosted model is separate and is labelled as such: if a workload has to stay in the country, Qwen 3.8 27B LK runs on our own server in Colombo.
Input is LKR 519 per 1M tokens; output is LKR 1,039 per 1M tokens. For example, 1,000 requests with 3,000 input and 1,000 output tokens each cost LKR 2,597. Rupee prices are our published US dollar rates converted at LKR 333 to the dollar, our rate for November 2026 — the same rate your invoice uses.
Yes. DeepSeek-V4-Pro usage is billed in LKR on one monthly invoice with every other model you use, so there is no foreign card or separate DeepSeek account to set up.
Its weights are open under the MIT licence, so you could run it yourself. Through Roar AI you call it like any other model, with no servers to manage.
DeepSeek-V4-Pro can read up to 1M tokens in one request, and write an answer of up to 384K tokens.
It reads text only. You can still send it a PDF: Roar AI turns each page into text first, at no extra charge. Scanned pages need a model that can see images.
Create a free account, get an API key and call DeepSeek-V4-Pro with the OpenAI SDK, billed in Sri Lankan rupees.