DeepSeek V3.2 API pricing
DeepSeek charges $0.28 per million input tokens and $0.4 per million output tokens for DeepSeek V3.2 (deepseek-v3.2), and $0.028 per million cached input tokens. A typical request of 1,000 input and 500 output tokens costs $0.00048.
DeepSeek V3.2 price per million tokens
| Token type | Standard |
|---|---|
| Input | $0.28 |
| Cached input | $0.028 |
| Output | $0.4 |
API model id: deepseek-v3.2. DeepSeek V3.2 supports a 164K-token context window, up to 164K output tokens, tool calling and reasoning.
What DeepSeek V3.2 costs per request
| Request | 1 request | 1,000 requests | 1M requests |
|---|---|---|---|
| Short chat message (300 in, 200 out) | $0.000164 | $0.16 | $164.00 |
| Typical request (1K in, 500 out) | $0.00048 | $0.48 | $480.00 |
| Long document summary (20K in, 1K out) | $0.006 | $6.00 | $6,000.00 |
| Coding agent step (50K in, 2K out) | $0.0148 | $14.80 | $14,800.00 |
At standard rates. A token is about 4 characters of English text, so 1,000 tokens is roughly 750 words. DeepSeek counts tokens with its own tokenizer, so the same text can be a different number of tokens.
How DeepSeek V3.2 compares
- For a typical request, DeepSeek V3.2 is cheaper than 80% of the other models on apicost.app.
- GPT-4o mini from OpenAI costs $0.15 input and $0.6 output per million tokens, so a typical request is $0.00045: about the same price as DeepSeek V3.2.
- Claude Haiku 5.5 from Anthropic costs $0.1 input and $0.5 output per million tokens, so a typical request is $0.00035: 1.4× cheaper than DeepSeek V3.2.
- Gemini 2.5 Flash Lite from Google Gemini costs $0.1 input and $0.4 output per million tokens, so a typical request is $0.0003: 1.6× cheaper than DeepSeek V3.2.
Models with a similar price
| Model | Provider | Input / 1M | Cached input / 1M | Output / 1M | Typical request |
|---|---|---|---|---|---|
| Gemini 2.5 Flash Lite | Google Gemini | $0.1 | $0.01 | $0.4 | $0.0003 |
| Ministral 3 14B 2512 | Mistral | $0.2 | $0.02 | $0.2 | $0.0003 |
| GPT-6 Luna | OpenAI | $0.1 | $0.01 | $0.5 | $0.00035 |
| Claude Haiku 5.5 | Anthropic | $0.1 | $0.01 | $0.5 | $0.00035 |
| Mistral Tiny | Mistral | $0.25 | $0.025 | $0.25 | $0.000375 |
| GPT-4o mini | OpenAI | $0.15 | $0.075 | $0.6 | $0.00045 |
| Open Mistral Nemo | Mistral | $0.3 | $0.03 | $0.3 | $0.00045 |
| Codestral 2508 | Mistral | $0.3 | $0.03 | $0.9 | $0.00075 |
Other DeepSeek models
- DeepSeek Coder $0.14 / $0.28
- DeepSeek Flash $0.3 / $1.2
- DeepSeek R1 $0.55 / $2.19
- DeepSeek V3 $0.27 / $1.1
- DeepSeek V4 Flash $0.3 / $1.2
- DeepSeek V4 Pro $1.32 / $3.96
Every DeepSeek model side by side, with batch, caching and long-context prices: DeepSeek API pricing. Ways to pay less: batch API, prompt caching, the cheapest models.
Questions and answers
How much does the DeepSeek V3.2 API cost?
DeepSeek charges $0.28 per million input tokens and $0.4 per million output tokens for DeepSeek V3.2 (deepseek-v3.2), and $0.028 per million cached input tokens. A typical request of 1,000 input and 500 output tokens costs $0.00048.
Is there a cheaper way to use DeepSeek V3.2?
Prompt caching lowers the price of input that repeats across requests to $0.028 per million tokens.
What is the context window of DeepSeek V3.2?
DeepSeek V3.2 accepts up to 163,840 input tokens, roughly 123,000 words of English, and writes up to 163,840 output tokens per response. Filling the whole context window once costs $0.0459 in input tokens.
Is DeepSeek V3.2 cheaper than GPT-4o mini?
They cost about the same. GPT-4o mini from OpenAI costs $0.15 input and $0.6 output per million tokens, so a typical request is $0.00045: about the same price as DeepSeek V3.2.