CRX id
cbhbgmdpcoelfdoihppookkijpmgahag
cbhbgmdpcoelfdoihppookkijpmgahag
Description from extension meta
Use your own AI models on the web
Image from store
Description from store
Window is the simplest way to connect AI models to the web. Use a variety of models for text generation and 3D creations (new!).
FOR USERS: bring your own models to the web, including ones running locally. Set up your models once, and use them everywhere.
FOR DEVELOPERS: no need to pay for OpenAI or other language model APIs. Make model-agnostic & multi-media AI apps. Users can try them for free.
Configure your models once; after that, a web app can request your permission to send a prompt to your chosen model. You get visibility on what's being asked and when.
Discover Window AI apps with live usage metrics on https://openrouter.ai.
FOR MODEL PROVIDERS: access more users on the web, without needing developers to change their code.
Supported text models:
- OpenAI's GPT-3.5, GPT-3.5 16k, and GPT-4
- Google's PaLM 2 Chat and Code Chat (Bison)
- Anthropic's Claude, Claude Instant, and Claude 100k
- Local models like Alpaca, Llama, and GPT4All. See https://www.localai.app/
- Together's GPT NeoXT 20B
- Cohere Xlarge
Supported 3D object models:
- OpenAI's Shape-E
Latest reviews
- (2023-08-27) Mikel: Started giving me unknown errors after working fine for a week. Can't figure out how to even get to their website to check an FAQ.
- (2023-05-31) Marissa Montgomery: Window AI is great!
- (2023-04-23) Jason Hu: This is dope!
- (2023-04-18) Nolan Clement: Building model-agnostic LLM apps is revolutionary and important. Additionally, it makes the developer experience for integrating large language models into apps easy and fast.
- (2023-04-18) David Pan: The user and developer experience is great! As a user, I can try many web apps without worrying about my API keys getting stolen, and can try different AI models. As a developer, I can reference the window.ai global variable to call the user's AI model.
- (2023-04-18) Yanni Kouloumbis: If we're to have an LLM-connected internet, we can’t expect people to be pasting their keys into insecure text fields, and we can’t expect developers to front all inference costs. Window AI is the first real attempt at solving this problem.
Latest issues
- (2023-09-13, v:0.0.21) P. M.: Learning to use window.ai
I don't yet get how this works. Will this work with AI chatbots one can train using re:Tune (retune.so)? If so, can you explain how? Thanks!