# DeepSeek V3 API Pricing

As of October 10, 2026, DeepSeek V3 costs $0.27 per 1M input tokens and $1.10 per 1M output tokens on DeepSeek, with cache reads at $0.07 and cache writes at $0 per 1M tokens. Across 15 providers on LiteLLM, the input price ranges from $0.20 to $1.50. It has a 65,536-token context window and returns up to 8,192 tokens per response.

- Page: https://models.litellm.ai/models/deepseek-v3
- Lab: DeepSeek
- LiteLLM model name: `deepseek/deepseek-v3`
- Context window: 65,536 tokens
- Max output: 8,192 tokens
- Added to LiteLLM: January 2025
- Prices updated: October 10, 2026

## Pricing by provider

Token prices in USD per 1M tokens.

| Provider | Model name | Input | Output | Cache read | Per image | Per second | Context |
|---|---|---|---|---|---|---|---|
| DeepSeek | `deepseek/deepseek-v3` | $0.27 | $1.10 | $0.07 | – | – | 64K |
| Amazon Bedrock | `deepseek.v3-v1:0` | $0.58 | $1.68 | – | – | – | 128K |
| Azure AI Foundry | `azure_ai/deepseek-v3` | $1.14 | $4.56 | – | – | – | 128K |
| Azure AI Foundry (snapshot 0324) | `azure_ai/deepseek-v3-0324` | $1.14 | $4.56 | – | – | – | – |
| Baseten (snapshot 0324) | `baseten/deepseek-ai/DeepSeek-V3-0324` | $0.77 | $0.77 | – | – | – | – |
| Crusoe (snapshot 0324) | `crusoe/deepseek-ai/DeepSeek-V3-0324` | $1.50 | $1.50 | – | – | – | 160K |
| DeepInfra | `deepinfra/deepseek-ai/DeepSeek-V3` | $0.32 | $0.89 | – | – | – | 160K |
| DeepInfra (snapshot 0324) | `deepinfra/deepseek-ai/DeepSeek-V3-0324` | $0.24 | $0.90 | $0.135 | – | – | 160K |
| Fireworks AI | `fireworks_ai/accounts/fireworks/models/deepseek-v3` | $0.90 | $0.90 | – | – | – | 128K |
| Fireworks AI (snapshot 0324) | `fireworks_ai/accounts/fireworks/models/deepseek-v3-0324` | $0.90 | $0.90 | – | – | – | 160K |
| GMI Cloud (snapshot 0324) | `gmi/deepseek-ai/DeepSeek-V3-0324` | $0.28 | $0.88 | – | – | – | 160K |
| Hyperbolic | `hyperbolic/deepseek-ai/DeepSeek-V3` | $0.20 | $0.20 | – | – | – | 32K |
| Hyperbolic (snapshot 0324) | `hyperbolic/deepseek-ai/DeepSeek-V3-0324` | $0.40 | $0.40 | – | – | – | 32K |
| Lambda (snapshot 0324) | `lambda_ai/deepseek-v3-0324` | $0.20 | $0.60 | – | – | – | 128K |
| Nebius AI Studio | `nebius/deepseek-ai/DeepSeek-V3` | $0.50 | $1.50 | – | – | – | 128K |
| Nebius AI Studio (snapshot 0324) | `nebius/deepseek-ai/DeepSeek-V3-0324` | $0.50 | $1.50 | – | – | – | 128K |
| Novita AI | `novita/deepseek/deepseek_v3` | $0.89 | $0.89 | – | – | – | 64K |
| Novita AI (snapshot 0324) | `novita/deepseek/deepseek-v3-0324` | $0.27 | $1.12 | $0.135 | – | – | 160K |
| Replicate | `replicate/deepseek-ai/deepseek-v3` | $1.45 | $1.45 | – | – | – | 64K |
| Together AI | `together_ai/deepseek-ai/DeepSeek-V3` | $1.25 | $1.25 | – | – | – | 64K |
| Vercel AI Gateway | `vercel_ai_gateway/deepseek/deepseek-v3` | $0.90 | $0.90 | – | – | – | 128K |

## Cost examples

1,000 requests with 2,000 input and 500 output tokens each:

- DeepSeek: $1.09
- Hyperbolic: $0.50
- Lambda: $0.70
- DeepInfra: $0.93

## Features

Supported: Function calling, Parallel tool calls, Structured output, Reasoning, Prompt caching.

## Use it with LiteLLM

```python
from litellm import completion

response = completion(
    model="deepseek/deepseek-v3",
    messages=[{"role": "user", "content": "Hello!"}],
)
```

## FAQ

### How much does DeepSeek V3 cost?

DeepSeek V3 costs $0.27 per 1M input tokens and $1.10 per 1M output tokens on DeepSeek, with cache reads at $0.07 and cache writes at $0 per 1M tokens. Across 15 providers on LiteLLM, the input price ranges from $0.20 to $1.50.

### What is the context window of DeepSeek V3?

DeepSeek V3 has a 65,536-token context window and returns up to 8,192 tokens per response.

### Which providers offer DeepSeek V3?

DeepSeek V3 is available from 15 providers through LiteLLM: DeepSeek, Amazon Bedrock, Azure AI Foundry, Baseten, Crusoe, DeepInfra, Fireworks AI, GMI Cloud, Hyperbolic, Lambda, Nebius AI Studio and Novita AI and 3 more. The lowest price is on Hyperbolic.

### How do I call DeepSeek V3 with an OpenAI-compatible API?

Use the LiteLLM Python SDK with model="deepseek/deepseek-v3", or add the model to the LiteLLM proxy's config.yaml and send requests to /v1/chat/completions from any OpenAI client. LiteLLM tracks the cost of every request at the prices on this page.

### What features does DeepSeek V3 support?

DeepSeek V3 supports function calling, parallel tool calls, structured output, reasoning and prompt caching.

### When did LiteLLM add DeepSeek V3?

DeepSeek V3 was added to LiteLLM's model price file in January 2025.

Source: LiteLLM model price file, https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json