You don't need to use their API. There's a github program called ChatAll that gives access to various models through API but also via web interfaces (you log into the service, whether that's Microsoft's copilot, higgingchat or perplexity, and you continue the chat in the ChatAll interface, not the web interface.
If we had access to such models in TypingMind, this would greatly increase it's use cases.
Amethyst Key
•
Apr 19, 2024
Came here to make this request as well and was happy to find that it is already being talked about.
Anyway, for the web version they historically used GPT-4 (now Turbo) for the Pro (former "Copilot") mode.
Tony Dinh
Team•
Mar 4, 2024
Diego Sala it's a little bit better but still very low quality to make a comparable experience as the current Web Search I think.
Silver Hubble
•
Mar 4, 2024
Tony Dinh, sure, the web version in GPT-4 Pro/Copilot mode is much better in quality. However, their API online models, as well as their web default version, still have some use for their very convenient speed and cost, especially considering that Pro users of Perplexity (like me) have $5/free credit for the API each month.
As for quality, I don't think there are better web-augmented API models than those of Perplexity, though. Do you?
Tony Dinh changed status to Planned
Team•
Mar 2, 2024
Yellow Flamingo
•
Feb 22, 2024
+1 for Perplexity search suggestions
Amaranth Dew
•
Feb 22, 2024
I was JUST about to make this feature request. Can this be done using the Perplexity API? If so we can use the existing settings from the custom model (for users who have added it)
Gray Crater
•
Feb 8, 2024
Yes, please add Perplexity Copilot for search, it is miles better than Google search.
Log in to comment and vote
Comments11
May 6, 2024
Maroon Telescope
Apr 21, 2024
You don't need to use their API. There's a github program called ChatAll that gives access to various models through API but also via web interfaces (you log into the service, whether that's Microsoft's copilot, higgingchat or perplexity, and you continue the chat in the ChatAll interface, not the web interface.
If we had access to such models in TypingMind, this would greatly increase it's use cases.
Amethyst Key
Apr 19, 2024
Came here to make this request as well and was happy to find that it is already being talked about.
Tony Dinh
Mar 2, 2024
Unfortunately, as of now Perplexity API is no where as good as their web interface. https://x.com/tdinh_me/status/1763911099211280655?s=20
Silver Hubble
Mar 2, 2024
Tony Dinh Try with their new model "sonar-medium-online" or their fine-tuned "llama-2-70b-chat", see here: https://docs.perplexity.ai/changelog/api-updates-february-2024
Pplx-70b-online is now deprecated.
Look at the attachment.
Anyway, for the web version they historically used GPT-4 (now Turbo) for the Pro (former "Copilot") mode.
Tony Dinh
Mar 4, 2024
Diego Sala it's a little bit better but still very low quality to make a comparable experience as the current Web Search I think.
Silver Hubble
Mar 4, 2024
Tony Dinh, sure, the web version in GPT-4 Pro/Copilot mode is much better in quality. However, their API online models, as well as their web default version, still have some use for their very convenient speed and cost, especially considering that Pro users of Perplexity (like me) have $5/free credit for the API each month.
As for quality, I don't think there are better web-augmented API models than those of Perplexity, though. Do you?
Mar 2, 2024
Yellow Flamingo
Feb 22, 2024
+1 for Perplexity search suggestions
Amaranth Dew
Feb 22, 2024
I was JUST about to make this feature request. Can this be done using the Perplexity API? If so we can use the existing settings from the custom model (for users who have added it)
Gray Crater
Feb 8, 2024
Yes, please add Perplexity Copilot for search, it is miles better than Google search.