Open AI NVDA Add-on: A Game-Changer for Accessibility

By Assistive Inteā€¦, 21 June, 2024

Forum
Windows

Iā€™m excited to share the incredible Open AI NVDA add-on, a powerful tool that integrates the capabilities of Open AI, Mistral, and OpenRouter into your NVDA workflow. Whether you need to craft detailed text, translate with precision, summarize documents, or interpret visual content, this add-on makes it effortless. Installation is straightforward, and configuring your API keys is a breeze. The main dialog and intuitive shortcuts enhance usability, making everyday tasks faster and more efficient. Kudos to the developer for creating such a useful, user-friendly tool. Check it out.

Options

Comments

By Gabriele Battaglia on Monday, June 24, 2024 - 00:19

Where is it?

By Gabriele Battaglia on Monday, June 24, 2024 - 00:19

Hello and thanks for your answer. I always dremt about this addon and when I read your message I was so thrilled I can't believe it. It seems terrific.
I'm a plus Open AI subscriber. I created my API Key and inserted it in the addon, then selected GPT 4o.
I sent a "Ciao" string and got an 404 error message.
I read the docs but didn't find any fixes for me. Any help please?
If you have the willing we can also contact directly by mail or WHatsApp, or facetime or Messenger.
Let me know.
Uh, maybe it is relevant: I'm in Italy.

By Assistive Inteā€¦ on Monday, June 24, 2024 - 00:19

I just wanted to give the developer some love and appreciation. I'm the same as you, i wanted somethig like this for ages.

I did exactly the same as you, tried the ChatGPT_ API key and it didn't work. I'm not very technical, so I just assumed it wasn't going to be that easy - that you would have to sign up for another Open AI API key, so that's what I did, for Open AI and Mistral.

I only had to pay something like $10 in advance and it has lasted for ages! I'm not planning on canceling the ChatGPT+ right away, I find it to useful.

I hope this helps, let me know if you need anymore. This is a great add-in, I love NVDA and how many amazing people have made so many amazing add-ins for it.

By Gabriele Battaglia on Monday, June 24, 2024 - 00:19

Thank you. I'll open an issue on github.

By Ollie on Monday, June 24, 2024 - 00:19

The API tokens are different to a chat GPT subscription. You need to add funding to your account on OPENAI.com to allow requests.

By Gabriele Battaglia on Monday, June 24, 2024 - 00:19

Thank you. Now it's clear. I though it was enough to pay for a Plus subscription, while they want more to use API keys.
Thanks again.

By Assistive Inteā€¦ on Monday, June 24, 2024 - 00:19

I think this add-in is ace! As I said, I wanted to spread the word and give the person who created it some love...sorry if I confused the issue. Maybe i should add the word 'review' if I do this again?

By Assistive Inteā€¦ on Wednesday, July 3, 2024 - 00:19

Looks like my desire to spread some love for this developer earnt them a slap on the wrists from Top Tech Titbits over there name! Sorry my friends, I'm sure you had the best of intentions and started when you were only offering Open AI access. as a big Mistral fan, I'm glad you went beyond.

Anyway, I just updated and got a message and a link to follow - here is a summary of what I found:

"Dear users of the OpenAI NVDA add-in,
Exciting news! The tool you've grown familiar with has evolved into BasiliskLLM, a fresh take on facilitating interactions with large language models (LLMs). Named after the infamous Roko's basilisk, this new incarnation promises enhanced usability and broader compatibility across various LLM platforms like Anthropic, Mistral, and OpenAI.
BasiliskLLM aims to be more than just functionalā€”it's designed to be intuitive enough for anyone, including screen reader users. Whether you're navigating conversations or exploring new features, this tool is crafted to blend convenience with reliability, ensuring a seamless experience in your interactions with LLMs.
For those accustomed to the NVDA add-on, transitioning to BasiliskLLM means embracing a tool that not only continues to support your needs but also opens doors to expanded functionalities and improved user interfaces. We're excited to embark on this journey with you as we refine and enhance BasiliskLLM together.
Download the latest version from our repository and experience the evolution firsthand. Your feedback and suggestions are invaluable as we strive to make BasiliskLLM your preferred choice in LLM interaction tools."

By blindpk on Wednesday, July 3, 2024 - 00:19

It works very much like the NVDA add-on. It has some quirks, but it is very accessible. It also have its own NVDA add-on, which is just a basic thing to make it possible to grab navigator objects or URLs from within NVDA. I think this app has potential, but we'll have to see how the development goes.

By Assistive Inteā€¦ on Wednesday, July 3, 2024 - 00:19

The add-in seems to stil work - I'm a bit put off by installing the other ting TBH - it seems very complicated.

By blindpk on Wednesday, July 3, 2024 - 00:19

As I understand it it just won't be developed anymore. I will continue to use the old addon until the app gets a bit better, but I think I'll switch then.
If you want the, in my opinion, easiest way, download the portable version of the app, place it in a folder and run it from there. If you don't like it, just remove the folder.

By Joshua on Wednesday, July 3, 2024 - 00:19

Can you shair a link to download it

Thanks

By Diego on Wednesday, July 3, 2024 - 00:19

Could anyone send the link to the latest portable version? Another question: Can I use some llm for free?

By blindpk on Thursday, October 10, 2024 - 00:19

BasiliskLLM has been steadily updated and yesterday the one big feature missing compared to the old NVDA addon was implemented, namely conversation profiles and, because of that, the ability to save system prompts etc.. Conversation profiles includes all parameters, model, system prompt, temperature and so on.
For those using NVDA and who have API keys for any of the supported AI services this app is really neat.
Currently there is a bug that hinders you from updating the application from within itself, but I guess that will be fixed soon. You can get the latest build here:
https://nightly.link/sigmanight/basiliskllm/workflows/ci/master

By Charlotte Joanne on Thursday, October 10, 2024 - 00:19

Is this a stand-alone app? Accessible I assume? Should I get it if I want to, for example, use Pixstral for images?

By blindpk on Thursday, October 10, 2024 - 00:19

Yes, it is standalone and it is fully accessible. It also includes a NVDA addon which makes it possible to send images or URLs to images from the current NVDA navigator object (you install the addon from within the app itself).
I'm not sure if it supports the Pixtral model, since I don't have a Mixtral API key, but yes, if it does this is. It is regurlarly updated with new models etc..

By Charlotte Joanne on Thursday, October 10, 2024 - 00:19

I'll give it a go. I do like being able to use a model directly.

And the one month I thought I had used it a lot, it was less than Ā£2!

By Charlotte Joanne on Thursday, October 10, 2024 - 00:19

Even though a lot of words I don't really understand came up, I managed to download it and complete setup for Mistral, no :ixstral. Now to collect some more API keys.

By Charlotte Joanne on Thursday, October 10, 2024 - 00:19

This is why I love NVDA. I've now got access to Antropic, Mistral and OPENAI. Just need to find a Gemini API key now.

By TheBllindGuy07 on Thursday, October 10, 2024 - 00:19

How do we install and use BasiliskLLM? I had a bad experience with it and I am still relying on the old openai addon?

By blindpk on Thursday, October 10, 2024 - 00:19

I'd say the easiest way is to go to https://github.com/sigmanight/basiliskllm/releases, under the 0.1a6 version (latest as of this writing) open "assets", download either the standard setup or the portable version (I would reccomend the setup package) and then install it.
Once you have the app installed, run it, configure one account, i.e. add an API key, and try it out. One tip, in the settings, you can configure how the app handles automatic updates. Consider changing the "channel" to "development" if you want all the latest features (and of course also the latest bugs), since the app is updated quite often and there has been a while between releases.