search

brianpetro / obsidian-smart-connections

Chat with your notes & see links to related content with AI embeddings. Use local models or 100+ via APIs like Claude, Gemini, ChatGPT & Llama 3
https://smartconnections.app
GNU General Public License v3.0
2.19k stars 155 forks source link

Request. Is there a way to add Gemini-API? #456

Open bbecausereasonss opened 4 months ago

bbecausereasonss commented 4 months ago

Not even sure if it's possible yet... but I just tested Gemini Ultra for 2 hours, and it absolutely BLEW my mind. It knocks ChatGPT out of the water on analysis, writing, and cognition.

brianpetro commented 4 months ago

@bbecausereasonss good to know! I'll have to look into it.

What were you using to test it?

🌴

bbecausereasonss commented 4 months ago

Just the web interface, a long with a bunch of reasoning questions. Work on branding, copy, and complex problems. Looking for creativity, comprehensive outlines, concept formulation, structure and cognition.

I came in skeptical but was very impressed.

Sent from Mobile.

On Thu, Feb 8, 2024, 7:07 PM WFH Brian @.***> wrote:

@bbecausereasonss https://github.com/bbecausereasonss good to know! I'll have to look into it.

What were you using to test it?

🌴

— Reply to this email directly, view it on GitHub https://github.com/brianpetro/obsidian-smart-connections/issues/456#issuecomment-1935122950, or unsubscribe https://github.com/notifications/unsubscribe-auth/AE2TIUHDOHMBWPRMVV2HLJDYSVSDXAVCNFSM6AAAAABDARMNECVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZVGEZDEOJVGA . You are receiving this because you were mentioned.Message ID: @.***>

antonioaraujo664 commented 4 months ago

I was also skeptical...it's better than chat gpt. I would also like to see Gemini in action.

alexmhoward3 commented 4 months ago

Second this. I’d love to see gemini’s 1 million token context window put to the test in my vault.

brianpetro commented 4 months ago

@alexmhoward3 the million context model still isn't generally available, but I'd also love to test it once it is!

bbecausereasonss commented 3 months ago

By the way, Claude 3 is also 'remarkably' good. Currently GPT4 is the weakest of the bunch. Of course OpenAI's embeddings still rock. Actually I believe Poe.com has an API which lets you use (any model)?

brianpetro commented 3 months ago

@bbecausereasonss v2.1 will go into beta/early release with supporters this week; it includes Claude and Gemini.

bbecausereasonss commented 3 months ago

@bbecausereasonss v2.1 will go into beta/early release with supporters this week; it includes Claude and Gemini.

Oooh, exciting!

alexmhoward3 commented 3 months ago

SUPER neat - thanks Brian! Quick q/possible bug. If I ask smart chat (using Claude 3) to reference a folder of notes, it hangs, and the stop button doesn't seem to do anything in the chat window. It looks like it outputs the response into a new note under a "smart-chats" folder, but the expected result is to see the response in the chat window. Screenshot 2024-03-13 at 1 20 44 PM

brianpetro commented 3 months ago

@alexmhoward3 thanks for the bug report! In a major release like v2.1, there are simply too many edge cases for me to test. So, I rely on reports like this 😊

The bug should now be fixed 🛠️

Next time you restart the plugin or Obsidian, the Smart Connections early-release version should automatically update to v2.1.3.

🌴

alexmhoward3 commented 3 months ago

@brianpetro confirmed fixed! thank you!