> ## 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-flash

> DeepSeek V4 Flash — small-mid tier of DeepSeek's V4 family. Anthropic and OpenAI-compatible. The best Chinese-LLM dollar-per-quality at the moment, well-suited to high-volume classification and agent loops.

`deepseek-v4-flash` is the **small-mid** of the DeepSeek V4 family. At \$0.14 / 1M input it undercuts Haiku and speaks both OpenAI Chat Completions and the Anthropic Messages shape — making it a popular cost-savings substitute for `gpt-5-4-mini` or `claude-haiku-4-5` in volume-sensitive pipelines.

**Pricing:** $0.14 / 1M input, $0.28 / 1M output, \$0.003 / 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` |

## Quickstart

<CodeGroup>
  ```bash cURL 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-flash",
      "messages": [
        { "role": "user", "content": "Classify: refund / billing / technical / other.\n\nMy invoice charged twice." }
      ]
    }'
  ```

  ```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-flash",
      messages=[{"role": "user", "content": "Classify this ticket."}],
  )

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

## Capabilities

| Capability         | Supported                 |
| ------------------ | ------------------------- |
| Chat Completions   | ✅                         |
| Anthropic Messages | ✅                         |
| Streaming (SSE)    | ✅                         |
| Tool use           | ✅                         |
| Parallel tools     | —                         |
| JSON mode          | ✅                         |
| Reasoning chain    | — (use `deepseek-v4-pro`) |
| Vision (HTTP API)  | ❌                         |
| Context window     | 64K tokens                |

## When to use

* **High-volume classification, routing, structured extraction** — cheapest Chinese-LLM tier with both OpenAI + Anthropic protocol coverage.
* **Agent loops on a tight budget** — `tool_use` works via the Anthropic endpoint at Haiku-comparable pricing.
* **Cost-optimized fallback** — when your task fits Flash, prefer this id directly over `deepseek-v4-pro` for \~3× cost savings.

When **not** to use:

* Tasks that need reasoning chain — go to `deepseek-v4-pro`.
* Vision input — not on HTTP API today.

## Next

* [deepseek-v4-pro](/models/deepseek-v4-pro) — reasoning flagship