> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cogito.decart.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# DeepSeek V4 Pro

> Frontier-class reasoning with a 1M-token context window.

DeepSeek's frontier model. 1M-token context, frontier-class reasoning, and a price tag that makes proprietary alternatives hard to justify.

|                |                                                                                            |
| -------------- | ------------------------------------------------------------------------------------------ |
| **Slug**       | `deepseek-v4-pro`                                                                          |
| **Parameters** | Frontier MoE                                                                               |
| **Context**    | 1,000,000 tokens (1M)                                                                      |
| **Throughput** | 70 tokens/sec                                                                              |
| **TTFT**       | 320ms                                                                                      |
| **License**    | DeepSeek License                                                                           |
| **Pricing**    | [cogito.decart.ai/models/deepseek-v4-pro](https://cogito.decart.ai/models/deepseek-v4-pro) |

## Best for

* Hardest reasoning, mathematical proofs, programming competitions
* Long-context analysis (full books, large codebases, extensive logs)
* Research workloads where you'd reach for a frontier closed-model API

```python theme={null}
client.chat.completions.create(
    model="deepseek-v4-pro",
    messages=[...],
)
```

## On context cost

The full 1M context is included in the standard token rate — no per-token surcharge for using the long window. Pair with [context caching](/getting-started/pricing#context-cache) and the economics on RAG workloads get very compelling.