Create an API key

DeepSeek V3.2 API: toman pricing and code

deepseek/deepseek-v3.2

Tool callingReasoningJSON output
Create a keyTry it
Input · per 1M tokens
90,017 toman
Output · per 1M tokens
131,563 toman
One 1,000-word request ≈
288 toman

You are billed for the usage the request actually reported. Prices follow the market. Cached input: 45,009 toman per 1M tokens.Pricing and top-ups

DeepSeek V3.2 example: tool calling

The example is picked from this model's own capabilities. Drop in your key and it runs as is.

from openai import OpenAI
import json

client = OpenAI(base_url="https://api.uttapen.ir/v1", api_key="sk-up-…")

tools = [{
    "type": "function",
    "function": {
        "name": "check_stock",
        "description": "Returns the stock level of a product",
        "parameters": {
            "type": "object",
            "properties": {"sku": {"type": "string"}},
            "required": ["sku"],
        },
    },
}]

messages = [{"role": "user", "content": "How many of the Nike NK-42 shoe are in stock?"}]
first = client.chat.completions.create(model="deepseek/deepseek-v3.2", messages=messages, tools=tools)
call = first.choices[0].message.tool_calls[0]

# you run the function yourself — the model never touches the database
args = json.loads(call.function.arguments)
result = {"sku": args["sku"], "qty": 7}

messages += [first.choices[0].message, {"role": "tool", "tool_call_id": call.id, "content": json.dumps(result)}]
final = client.chat.completions.create(model="deepseek/deepseek-v3.2", messages=messages, tools=tools)
print(final.choices[0].message.content)
import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://api.uttapen.ir/v1", apiKey: "sk-up-…" });

const tools = [{
  type: "function",
  function: {
    name: "check_stock",
    description: "Returns the stock level of a product",
    parameters: { type: "object", properties: { sku: { type: "string" } }, required: ["sku"] },
  },
}];

const messages = [{ role: "user", content: "How many of the Nike NK-42 shoe are in stock?" }];
const first = await client.chat.completions.create({ model: "deepseek/deepseek-v3.2", messages, tools });
const call = first.choices[0].message.tool_calls[0];

const args = JSON.parse(call.function.arguments);
const result = { sku: args.sku, qty: 7 };

messages.push(first.choices[0].message, { role: "tool", tool_call_id: call.id, content: JSON.stringify(result) });
const final = await client.chat.completions.create({ model: "deepseek/deepseek-v3.2", messages, tools });
console.log(final.choices[0].message.content);
curl https://api.uttapen.ir/v1/chat/completions \
  -H "Authorization: Bearer sk-up-…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v3.2",
    "messages": [{"role": "user", "content": "How many of the Nike NK-42 shoe are in stock?"}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "check_stock",
        "parameters": {"type": "object", "properties": {"sku": {"type": "string"}}, "required": ["sku"]}
      }
    }]
  }'

What is DeepSeek V3.2 good for?

The id for DeepSeek V3.2 in our API is "deepseek/deepseek-v3.2", served from DeepSeek. Context is 163,840 tokens; past that you have to summarise the history yourself. A single response can run to 147,456 tokens. On price it sits in the "very cheap" band — cheaper than 325 and dearer than 73 of the other paid models in the catalogue.

Capabilities available on this id: it supports tool calling, so it can invoke your own functions with valid arguments; it returns schema-valid JSON through response_format, ready to hand to your code; it has a reasoning mode that pays off on multi-step problems, maths and debugging. All of it works through the standard parameters of the official OpenAI SDK — no custom client, no wrapper. Keep in mind that reasoning tokens are output tokens and do appear on the bill.

For a back-of-envelope figure: about 288 toman per 1,000-word exchange (90,017 in, 131,563 out, per 1M tokens). It supports cached input: repeated context is billed at 45,009 toman per 1M, which matters a lot if your system prompt is long. You are always charged for the usage the request actually reported, never for the estimate, and a failed request costs nothing.

The closest alternative with the same capabilities from a different provider is Qwen2.5 72B Instruct: DeepSeek V3.2 works out about 17% cheaper, and its context window is larger. Both run on the same key and the same code, so trying the other one is a single string change.

To make the figure concrete: 100,000 toman of credit buys roughly 347 thousand-word requests on DeepSeek V3.2, and every 1,000 toman is about 3,472 words of round trip. A job with one million input tokens and one million output tokens comes to 221,580 toman in total. Filling this model's 163,840-token window costs 14,748 toman on the input side alone, which is the real reason to keep conversation history short.

Its nearest relative in the catalogue is "deepseek/deepseek-v3.2-exp", and choosing between those two is where most people hesitate. Both land at nearly the same price on a thousand-word request (288 and 306 toman). On parameters, this one takes logit_bias, logprobs, min_p, presence_penalty.

DeepSeek has 18 models in our catalogue; the cheapest is DeepSeek V4 Flash 0423 at 123 toman per thousand words and the dearest DeepSeek Pro Latest at 3,642. Among the less common parameters it accepts logit_bias, logprobs, min_p, presence_penalty, repetition_penalty, top_k — all through the standard request body. By context size the nearest option from another provider is Llama Guard 4 12B at 163,840 tokens.

What to use it for: high-volume work such as classification, tagging and bulk summarising, logic puzzles and code review, agents that reach out to APIs and databases, extracting data against a fixed schema.

Provider's own description

DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism...

Three real jobs, priced on this model

Each figure is derived from the prices above and moves when they do.

JobTokensCost
One chat turn with a medium history1,500 in + 400 out188 toman
Summarising a ten-page document4,000 in + 600 out439 toman
Classifying a thousand short rows120,000 in + 20,000 out13,433 toman

Price history

DateInput (toman/1M)Output (toman/1M)
31 Oct 202690,017131,563
22 Oct 202696,941145,412
22 Oct 202624,23536,353
17 Oct 202696,941145,412
10 Oct 202689,671276,975
8 Oct 202689,671145,412
6 Oct 202696,941145,412
5 Oct 202689,671145,412
28 Sep 202696,941145,412
7 Sep 202693,133138,488

Every price change for this model is recorded. Toman figures use today's rate.

Frequently asked

How do I call DeepSeek V3.2 from Iran?
Sign up with your mobile number, top the wallet up in toman, create an API key, then in the official OpenAI SDK point base_url at https://api.uttapen.ir/v1 and set model to "deepseek/deepseek-v3.2". Nothing else in your code changes.
What does DeepSeek V3.2 cost in toman?
90,017 toman per 1M input tokens and 131,563 toman per 1M output tokens; a 1,000-word request is around 288 toman. You pay for the usage that request actually reported, and a failed request costs nothing.
How much input does DeepSeek V3.2 take?
Up to 163,840 tokens per request, roughly 123k words. A single answer can reach 147,456 tokens.
Does DeepSeek V3.2 support streaming and tool calling?
Streaming (stream=true) works on every model here. This one supports tool calling in the standard OpenAI shape. Structured output through response_format works too.