GPT-API-free / DeepSeek-API-free

Free use of gpt-5 | deepseek

Supports gpt | deepseek | claude | gemini | grok

Domestic dynamic acceleration | Direct connection without proxy | Unified protocol for easy integration

Quick Start / API Documentation / Apply for Internal Beta Free Key / Support Paid Key / Service Availability

QQ Group: 1075658240

Privacy Statement

This project highly values privacy and is committed to protecting the privacy of its users. The project does not collect, record, or store any text input by users or any text returned by OpenAI servers in any way. The project does not provide OpenAI or any third party with any information regarding the identity of the API caller, including but not limited to IP addresses and user agent strings.

However, OpenAI officials will retain data for 30 days in accordance with their Data Usage Policy.

Features

1. Supports top-ranked common large models such as gpt | deepseek | claude | gemini | grok.
2. Free version support:
   • gpt-5.2, gpt-5.1, gpt-5, gpt-4o, and gpt-4.1: 5 times per day.
   • deepseek-r1, deepseek-v3, deepseek-v3-2-exp: 30 times per day.
   • gpt-4o-mini, gpt-3.5-turbo, gpt-4.1-mini, gpt-4.1-nano, gpt-5-mini, gpt-5-nano: 200 times per day.
3. Interface standards completely consistent with the official version, compatible with various software/plugins.
4. Supports streaming responses.
5. Domestic routes use dynamic acceleration; the experience is significantly better than using a proxy to connect to official endpoints.
6. No need for specialized internet access (VPN); directly available in domestic environments.
7. Completely free for personal use.
8. Protocols uniformly use the OpenAI standard protocol; for models from other vendors, only the model name needs to be changed, making integration convenient.

🚩 Important Notes

❗️If you encounter issues like no response or errors, check status.chatanywhere.tech to confirm the service status and assist in troubleshooting.

❗️The reasoning capabilities of the Free API Key gpt-5 series models are relatively weak; if stronger reasoning is required, please purchase a paid API.

❗️Free API Keys are strictly for personal non-commercial use, education, and non-profit scientific research. Commercial use is strictly prohibited, as is the large-scale training of commercial models! Contact us via the group in advance for research model training.

❗️We will periodically ban keys that are found to be abused. If you find your key has been mistakenly banned, please contact us via the QQ group.

❗️Our system is for internal evaluation and testing only; use for commercial or public-facing purposes is at your own risk.

To ensure long-term development, Free API Keys are limited to 200 requests/day/IP & Key (GPT and Embeddings are calculated separately at 200 each). This means if you use multiple keys under one IP, the total daily requests for all keys cannot exceed 200; similarly, if one key is used across multiple IPs, the daily requests for that key cannot exceed 200. (Paid version APIs do not have this limit.)

Free Use

• 🚀 Apply for Internal Beta Free API Key
• Free version supports deepseek, gpt-3.5-turbo, embedding, gpt-4o series, and gpt-5 series.
• Forwarding Host 1: https://api.chatanywhere.tech (Domestic transit, lower latency)
• Forwarding Host 2: https://api.chatanywhere.org (For international use)

We will expand capacity regularly based on usage. As long as we are not sanctioned by official providers, we will continue to provide free APIs. If this project helps you, please give us a Star. Feedback can be provided in Issues.

To use the key for forwarding, change the Host to api.chatanywhere.tech (preferred for domestic) or api.chatanywhere.org (international).

Paid Version API

• Providing free keys is purely a public welfare effort and not a sustainable long-term solution. Therefore, we introduced paid API keys to maintain daily expenses and promote a virtuous cycle for the project.
• Purchase Paid Key
• Paid Version Price List

1. Supports a more stable and faster GPT4 API with a better experience, unlimited use, lower prices than official rates, and convenient recharging.
2. Follows the official billing strategy: streaming Q&A uses the tiktoken library for accurate token calculation; non-streaming uses official returned token counts.
3. Balance never expires. Based on feedback, 30 RMB for moderate personal use of gpt-4o-mini can last roughly half a year.
4. All interfaces (including free versions) are guaranteed to be forwarded from official OpenAI or Azure interfaces—no unstable "peo" or "plus" solutions or reverse engineering—ensuring stability.

Paid Version Supported Models

(Note: Pricing is in CA/1K Tokens unless otherwise specified)

(Table abridged for readability; all models listed in original file are supported)

How to Use

• Due to frequent malicious requests, we no longer provide public free keys directly. You must bind your GitHub account to receive your own free key.
• 🚀 Apply for Internal Beta Free API Key or Purchase Paid API Key
• Forwarding Host 1: https://api.chatanywhere.tech (Domestic transit, lower latency)
• Forwarding Host 2: https://api.chatanywhere.org (For international use)
• Balance & Usage Inquiry: Balance & Announcements
• Forwarding APIs cannot request api.openai.com directly; change the request address to api.chatanywhere.tech.

Usage in Common Software/Plugins

Python OpenAI Official Library

Sample code can be found in demo.py.

Method 1

from openai import OpenAI

client = OpenAI(
    api_key="YOUR API KEY",
    base_url="[https://api.chatanywhere.tech/v1](https://api.chatanywhere.tech/v1)"
)