Deepseek

deepseek-v4-flash

Auto-router

deepseek/deepseek-v4-flash

ChatToolsSystem promptReasoning

Context

Max output

348K

Input / 1M

$0.14

Output / 1M

$0.28

Cached / 1M

$0.0028

Cutoff

May 2024

About this model

DeepSeek-V4-Flash is a fast, capable model supporting thinking and non-thinking modes with 1M context. Accepts text inputs and produces text outputs.

Best suited for

Multi-turn conversation, reasoning tasks, JSON output, function/tool calling, and high-throughput pipelines.

Built-in tools

Gtwy web search

Hosted by the gateway — enable them per request without wiring your own endpoint.

Capabilities

Tools

Native function calling, so agents can invoke your endpoints.

System prompt

Honours a dedicated system role, separate from the user turn.

Reasoning

Emits a separate thinking pass before the answer.

Supported parameters

stop

This parameter tells the model to stop generating text when it reaches any of the specified sequences (like a word or punctuation)

creativity_level

Controls randomness. Higher values (0.8) make output more random, lower values (0.2) more focused and deterministic.

probability_cutoff

Nucleus sampling parameter. The model considers tokens with top_p probability mass.

log_probabilityLog Probabilities

Whether to return log probabilities of the output tokens.

max_tokensMax Tokens Limit

Specifies the maximum number of text units (tokens) allowed in a response, limiting its length.

tools

Lists tool definitions or capabilities available to the model.

tool_choice

Decides whether to use tools or just the model for generating responses.

response_type

Defines the format or type of the generated response.

parallel_tool_calls

Enables parallel execution of tools, allowing multiple tools to run simultaneously.

reasoning

Controls the level of reasoning used by the model.

stream

Sends the response in real-time as it's being generated.

service_tier