DeepSeek logo

Free DeepSeek API Key, Base URL & Rate Limits

Model Provider

DeepSeek provides direct API access to its flagship models — DeepSeek-Chat (V3.

Function Calling
Save API Key Test API Key Get API Key Docs

How to get a free DeepSeek API key

  1. 1
    Sign up at platform.deepseek.com Email or phone registration. No credit card.
  2. 2
    Go to API Keys in the dashboard
  3. 3
    Create a new API key
  4. 4
    Choose a model deepseek-chat (V3.2) for general use. deepseek-reasoner (R1) for complex reasoning.
  5. 5
    Configure OpenAI client Base URL: https://api.deepseek.com/v1. Fully OpenAI SDK-compatible.

Provider Snapshot

P Provider Type Model Provider
O OpenAI Compatible No
U Base URL https://api.deepseek.com/v1
F Free Tier Check provider
P Phone Required No
S Streaming Varies
T Function Calling Yes
V Vision Varies
L Last Updated 2026-06-15

Supported Models 2 models

View in directory →
Model ID Developer Context Availability Free Tier Use cases
deepseek-chat-v3-2
deepseek-chat (V3.2)
 DeepSeek 128K Check provider Yes chat
deepseek-reasoner-r1
deepseek-reasoner (R1)
 DeepSeek 128K Check provider Yes chatreasoning

Developer Tools

K Save API Key Never lose your API keys — save keys from all providers in one place. T Test API Key Test your API key and check quota.

Related Providers

View all →
OpenRouter Chutes.ai Glhf.chat Grok (xAI) Groq

DeepSeek FreeLLM Score free API access score

How we score →
38 /100
🔹 Niche Provider — Consider for easy signup
Generosity Free limits and access terms
65
Access Signup and key availability
75
Model Breadth Free model coverage
0
Reliability Status and stability signals
15
Compatibility SDK and endpoint support
15
Quality Model capability signals
60

What is DeepSeek?

DeepSeek's official API — V3.2 + R1 with 128K context, OpenAI-compatible.

DeepSeek provides direct API access to its flagship models — DeepSeek-Chat (V3.2) and DeepSeek-Reasoner (R1) — through its own platform at platform.deepseek.com. The API is fully OpenAI-compatible, making it a drop-in replacement. Free tier has dynamic rate limits based on usage patterns. No credit card required for registration.

  • DeepSeek V3.2 (chat) + R1 (reasoning)
  • 128K context window
  • Fully OpenAI-compatible endpoint
  • No credit card required

API Compatibility: Live-tested OpenAI-compatible Chat Completions

DeepSeek Free Tier Limits & Pricing

Credit Card Not required
Phone Verification Not required
Free Tier Check provider
Context Range 128K
Total Models 2 listed
Rate Limits Dynamic
API Compatibility Live-tested OpenAI-compatible Chat Completions

DeepSeek API Setup Tutorials

Claude Code Environment variables and base URL setup Codex OpenAI-compatible config for local coding Gemini CLI Provider-specific API key routing notes

DeepSeek is fully compatible with popular AI coding assistants like Cursor, Claude Code, and more. To see step-by-step API configuration instructions for your favorite tool, please visit our Global Configuration Guide →

DeepSeek Model Use Cases

What DeepSeek's free models are best for, based on aggregated model capabilities:

Chat 2 models Reasoning 1 model

DeepSeek Limitations & Caveats

  • Dynamic rate limits — unpublished, adjusts based on usage
  • Only 2 models available directly
  • DeepSeek models also available on OpenRouter and NVIDIA NIM (with different limits)

DeepSeek FAQ

Why use DeepSeek's own API instead of OpenRouter or NVIDIA NIM?

DeepSeek's direct API has two advantages: 1) potentially higher rate limits for heavy users (dynamic scaling), 2) always has the latest model version first. OpenRouter and NVIDIA NIM offer the same models but with fixed (often lower) rate limits.

What's the difference between DeepSeek-Chat (V3.2) and DeepSeek-Reasoner (R1)?

DeepSeek-Chat (V3.2) is the general-purpose chat model — faster, cheaper, good for most tasks. DeepSeek-Reasoner (R1) uses chain-of-thought reasoning and excels at math, logic, and complex problem-solving — but is slower and uses more tokens.

Does DeepSeek's dynamic rate limit mean it's unlimited?

No — dynamic means the limit adjusts based on your usage patterns and current server load. New users typically get lower limits that increase with consistent, non-abusive usage. For predictable limits, use OpenRouter (200 RPD) or NVIDIA NIM (40 RPM).