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

# deepseek-v4-pro

> DeepSeek V4 Pro — the cost-effective reasoning flagship. A fraction of the cost of gpt-5-4 at comparable quality on most benchmarks. Dual endpoint (Chat Completions + Anthropic Messages). Reasoning chain exposed. Vision not on HTTP API.

`deepseek-v4-pro` is DeepSeek's reasoning flagship and the most cost-effective high-capability model on ByteSpike. A fraction of the cost of `gpt-5-4` at **comparable quality on most benchmarks**, with the same dual-protocol surface that DeepSeek exposes natively.

**Pricing:** $0.435 / 1M input, $0.87 / 1M output, \$0.004 / 1M cache read — see the [rate card](/pricing).

## Protocols

| Protocol                | Path                                                |
| ----------------------- | --------------------------------------------------- |
| Anthropic Messages      | `POST https://llm.bytespike.ai/v1/messages`         |
| OpenAI Chat Completions | `POST https://llm.bytespike.ai/v1/chat/completions` |

Both protocols are first-class. The same ByteSpike key works on either; pick whichever shape your client already speaks.

## Quickstart

<CodeGroup>
  ```bash cURL (OpenAI shape) theme={null}
  curl https://llm.bytespike.ai/v1/chat/completions \
    -H "Authorization: Bearer $BYTESPIKE_API_KEY" \
    -H "content-type: application/json" \
    -d '{
      "model": "deepseek-v4-pro",
      "messages": [
        { "role": "user", "content": "Hello, ByteSpike." }
      ]
    }'
  ```

  ```bash cURL (Anthropic shape) theme={null}
  curl https://llm.bytespike.ai/v1/messages \
    -H "x-api-key: $BYTESPIKE_API_KEY" \
    -H "anthropic-version: 2023-06-01" \
    -H "content-type: application/json" \
    -d '{
      "model": "deepseek-v4-pro",
      "max_tokens": 1024,
      "messages": [
        { "role": "user", "content": "Hello, ByteSpike." }
      ]
    }'
  ```

  ```python Python (openai SDK) theme={null}
  from openai import OpenAI

  client = OpenAI(
      base_url="https://llm.bytespike.ai/v1",
      api_key=os.environ["BYTESPIKE_API_KEY"],
  )

  resp = client.chat.completions.create(
      model="deepseek-v4-pro",
      messages=[{"role": "user", "content": "Hello, ByteSpike."}],
  )

  print(resp.choices[0].message.content)
  ```
</CodeGroup>

## Capabilities

| Capability                  | Supported                                        |
| --------------------------- | ------------------------------------------------ |
| Chat Completions            | ✅                                                |
| Anthropic Messages          | ✅                                                |
| Streaming (SSE)             | ✅                                                |
| Tool use (function calling) | ✅ parallel                                       |
| JSON mode                   | ✅                                                |
| **Reasoning chain**         | ✅ (`reasoning_content` field / `thinking` block) |
| Vision (HTTP API)           | ❌                                                |
| Context window              | 64K tokens                                       |

The reasoning chain ("thinking output") appears as:

* On the OpenAI endpoint: `reasoning_content` field on the choice.
* On the Anthropic endpoint: a `thinking` block ahead of the `text` block in `content[]`.

Reasoning tokens count toward output billing.

## When to use

* **Cost-sensitive reasoning** — when you'd reach for `gpt-5-4` but want a fraction of the cost.
* **Agents on a budget** — `tool_use` blocks pass through transparently via the Anthropic endpoint; DOSIA Agent uses this model heavily.
* **DOSIA Agent mode** — must use the Anthropic Messages endpoint.

When **not** to use:

* Vision-required tasks — not on HTTP API today.
* Web search — DeepSeek does not expose a grounding tool.
* Long context — 64K is the ceiling here; go to `claude-opus-4-8` (200K) or `gemini-3-5-flash` (1M).

## Next

* [deepseek-v4-flash](/models/deepseek-v4-flash) — small-mid