# DeepSeek-V4-Flash (0731)

> DeepSeek's sparse mixture-of-experts revision for coding, reasoning and agent workflows, with a 1M-token context window.

## Key facts

- Model author: DeepSeek
- Input: $0.4286 per 1M tokens (starting tier)
- Output: $1.2857 per 1M tokens (starting tier)
- Cached input: $0.0429 per 1M tokens
- Context window: 1,000,000 tokens
- Maximum output: 384,000 tokens
- Input modalities: text
- Output modalities: text
- Capabilities: mcp, prompt_caching, reasoning, tools, web_search
- Released: 2026-04-24
- Model IDs: `baidu/deepseek-v4-flash-0731`, `jd/deepseek-v4-flash-0731`
- Model page: https://www.realrelay.ai/model/deepseek/deepseek-v4-flash-0731

## Overview

DeepSeek-V4-Flash (0731) is a sparse mixture-of-experts model from DeepSeek. It is a post-training revision suited to coding, reasoning and agent workflows. It supports thinking and non-thinking modes, tool calling and web search, takes text input and returns text, and offers a very large context window.

DeepSeek-V4-Flash (0731) has a 1M-token context window and can generate up to 384K output tokens. It accepts text input and returns text output. It was released on April 24, 2026.

## Capabilities

DeepSeek-V4-Flash (0731) supports reasoning, tool calling, MCP, prompt caching and web search. It accepts text as input and produces text as output.

Thinking mode is the default, and non-thinking mode is available when quick answers matter more than reasoning. The model also supports tool calling, JSON output, MCP and prompt caching, which fits agent loops that call functions and reuse long prompts.

## Pricing

DeepSeek-V4-Flash (0731) starts from $0.43 per 1M input tokens and from $1.29 per 1M output tokens. Cache reads start from $0.043 per 1M tokens.

Pricing is dynamic, so the prices shown are starting prices. DeepSeek-V4-Flash (0731) is available through RealRelay's OpenAI-compatible API.

## Use cases

Use DeepSeek-V4-Flash (0731) for coding assistants and agents that call tools: its 1M-token window holds a large codebase, and cached prompt prefixes are priced lower than fresh input.

Its thinking mode also suits reasoning-heavy tasks such as analysing long documents, while non-thinking mode keeps simple requests quick.

## Call it

OpenAI-compatible chat completions:

```bash
curl https://www.realrelay.ai/v1/chat/completions \
  -H "Authorization: Bearer $REALRELAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "jd/deepseek-v4-flash-0731", "messages": [{"role": "user", "content": "Hello"}]}'
```

## FAQ

### What is DeepSeek-V4-Flash (0731)?

DeepSeek-V4-Flash (0731) is a sparse mixture-of-experts model from DeepSeek. It supports reasoning, tool calling, MCP, prompt caching and web search and accepts text input.

### How much does DeepSeek-V4-Flash (0731) cost?

DeepSeek-V4-Flash (0731) starts from $0.43 per 1M input tokens and from $1.29 per 1M output tokens, with cache reads from $0.043 per 1M tokens. Pricing is dynamic, so these are starting prices.

### What is the context window of DeepSeek-V4-Flash (0731)?

DeepSeek-V4-Flash (0731) has a 1M-token context window and can generate up to 384K tokens in a single response.

### What inputs and outputs does DeepSeek-V4-Flash (0731) support?

DeepSeek-V4-Flash (0731) accepts text input and returns text output.

### Does DeepSeek-V4-Flash (0731) support tool calling?

Yes. DeepSeek-V4-Flash (0731) supports tool calling, so it can call functions you define and use the results in multi-step tasks. It also supports reasoning, MCP, prompt caching and web search.

### When was DeepSeek-V4-Flash (0731) released?

DeepSeek-V4-Flash (0731) was released on April 24, 2026.

### How do I call DeepSeek-V4-Flash (0731) via the API?

Call DeepSeek-V4-Flash (0731) through RealRelay's OpenAI-compatible API using the model ID jd/deepseek-v4-flash-0731.