Update date: 20230410
Domestic ChatGPT tools
Yuan Cheng Xiang ChatImg
Developed by a team led by Lu Zhiwu, a professor at the Hillhouse School of Artificial Intelligence at Renmin University of China.
You can read pictures and chat, you need to register with a mobile phone number, and you can use it after registration.
Wenxinyiyan
It needs to be applied before it can be used. I applied on March 17th and passed it on April 6th. I waited for a total of 20 days.
The programming effect is much better than that of ChatGPT, and I can’t understand the meaning of the question clearly. I feel like I’m counting words when I talk back and forth.
ChatGPT mirror site
website
[Game Ai] ChatGPT Unofficial Domestic Mirror (bo-e.com)
Error: [[OpenAI] Server was limited, please try again later]
Only for development and learning exchange (binjie.site)
These mirror sites currently do not require an api-key to be used
Intelligent chat robot (123bot.cn)
You need to log in with WeChat to use it, and you will be prompted if you don’t log in
The current number of visitors is too large, if the robot does not respond
Please exit and re-enter, or click the retry button [ 403 ]
For more mirror sites, you can go to the following URL to find
Mobile terminal
WeChat public account
【Intelligent Chat AI Robot】
Yuan Chengxiang
Free ChatGPT api-key
Unicorn Sharing: Daily updated free ChatGPT account, free OpenAIkey (freeopenai.xyz)
According to the introduction, share more than 15 free keys and give 4 chatGPT accounts every day
browser plug-in
Checkget AI
Checkget AI - Smart Assistant for GPT-4 - Chrome Web Store (google.com)
ChatGPT Assistant
Merlin - ChatGPT Assistant for All Websites - Chrome Web Store (google.com)
Open source local deployment ChatGPT tool
ChatGLM
Zhipu AI, a company transformed from Tsinghua’s technological achievements, announced a new member of the open-source GLM series of models—the Chinese-English bilingual dialogue model ChatGLM-6B, which supports reasoning on a single consumer-grade graphics card
Vicuna vicuna
lm-sys/FastChat: The release repo for “Vicuna: An Open Chatbot Impressing GPT-4” (github.com)
Stanford scholars teamed up with CMU, UC Berkeley, etc. to launch a new model again - Vicuna with 13 billion parameters, commonly known as "vicuna" (llama), which is said to have reached 90% of chatgpt capabilities, and deployment requires at least 8GB of GPU memory, relying on CPU 60GB RAM required to run
demo address: FastChat (lmsys.org)
Asking questions on the demo, the response speed is very slow
LLaMA
The large-scale language model LLaMA released by Meta, with only 1/10 of the parameter LLaMA (13 billion), surpasses GPT-3 under most benchmark tests, and runs the LLaMA large model on a notebook
ggerganov/llama.cpp: Port of Facebook’s LLaMA model in C/C++ (github.com)