Text · DeepSeek

DeepSeek

Reasoning-first LLMs via RunAPI — flash for fast, low-cost work; pro for complex agentic tasks.

Operational · 2 variants · from $0.060

# Works with Claude Code, Codex, Gemini CLI, Cursor, and 50+ agents
npx skills add runapi-ai/deepseek -g

The -g flag installs globally so every project picks it up.

Or paste this prompt to your AI agent:

Install the DeepSeek skill for me:

1. Clone https://github.com/runapi-ai/deepseek
2. Copy the skills/deepseek/ directory into your
   user-level skills directory (e.g. ~/.claude/skills/
   for Claude Code, ~/.codex/skills/ for Codex).
3. Verify that SKILL.md is present.
4. Confirm the install path when done.

OVERVIEW

DeepSeek is a family of reasoning-first language models. deepseek-v4-flash is the fast, low-cost tier with an optional thinking mode; deepseek-v4-pro targets the hardest reasoning and agentic workloads. Both are available through RunAPI with one key and per-token billing.

Multiple variants for different speed / quality tiers
Single endpoint per action — text-to-X, image-to-X, etc.
Streaming and async patterns supported per variant
Failed generations are not charged

VARIANTS

Compare all variants

Variant	Billing	From
deepseek-v4-flash	1K tokens	$0.060	View →
deepseek-v4-pro	1K tokens	$0.500	View →

AGENT USAGE

From prompt to tool call

Typed schema · validates before the call hits the wire

# User prompt to the agent
"Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."

// Code generated by the agent via @runapi.ai/deepseek
import { DeepseekClient } from '@runapi.ai/deepseek';

const client = new DeepseekClient();
const result = await client.message.run({
  model: 'deepseek-v4-flash',
  prompt: 'Refactor this Python module for readability, explain each change, then add unit tests for the edge cases.',
});

@runapi.ai/deepseek npm

HOW IT WORKS

From install to first result in four steps

Install

One CLI command adds the skill bundle to your agent.

Configure

Set the variant ID and your RunAPI key. That's the whole config.

Call

The agent emits a typed tool call. Schema-validated, no glue code.

Receive

Stream the result back into your loop. Cached, signed, ready.

SDK QUICKSTART

Call directly from your code

runapi.ai

curl -X POST https://runapi.ai/v1/messages \
  -H "Authorization: Bearer $RUNAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "deepseek-v4-flash",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."
    }
  ]
}'

import { DeepseekClient } from "@runapi.ai/deepseek";

const client = new DeepseekClient();
const result = await client.message.run({
    model: "deepseek-v4-flash",
    max_tokens: 1024,
    messages: [{"role":"user","content":"Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."}],
});

require "runapi/deepseek"

client = RunApi::Deepseek::Client.new
result = client.message.run(
    model: "deepseek-v4-flash",
    max_tokens: 1024,
    messages: [{role: "user", content: "Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."}]
)

@runapi.ai/deepseek v1

CONTEXT

What is DeepSeek?

DeepSeek models are large language models tuned for reasoning, code, and agentic tool use. Through RunAPI they share a single API key with pay-as-you-go token billing, and are callable from both the OpenAI and Anthropic SDKs.

Provider

DeepSeek

Modality

Text

WHY RUNAPI