@perplexity_ai Honestly, I feel like the way Perplexity treats Pro users is getting increasingly awful. Can you just scrap the ‘Pro’ plan altogether and only keep Max? Or is ‘Pro’ still your main revenue source but you treat it like a stepchild?
@perplexity_ai After all, Perplexity’s Deep Research is still the best reasoning-based search model. Too bad they don’t offer it as a standalone model like the others they integrate. Somehow, its accuracy drops after a few uses — unlimited in quantity, but very limited in quality.
@Nick_Davidov Perplexity’s models are mainly for search. They’re good, but to really experience the model, use the official one. For pure reasoning or coding work, the original model is best.
@AravSrinivas@comet Overall, user interactions such as operations, movements, and opening/closing tabs are not as smooth as they should be, especially when compared to the Perplexity app designed for iOS.
@AravSrinivas@comet Suddenly causes the entire system to lag, especially Comet itself when it is ‘thinking’ and performing automated actions, but the user switches to another tab while waiting, such as to YouTube, etc. Restarting the PC is the only way to really stop it.
@jeffgrimes9 It’s nothing new or unique either. Personally, I think the feature that suggests buy/sell points as well as the timing to buy/sell using AI technology is what Perplexity should really focus on
@perplexity_ai The model is too slow and still can’t outperform Gemini 2.5 Pro. Also, any model integrated with Perplexity becomes less accurate because it depends on Perplexity’s search input.
@ionathanalcant I am surprised to discover that the accuracy of ChatGPT’s search function, when coupled with its reasoning process, now exceeds that of Perplexity Pro, even when the latter utilizes all available models.
@testingcatalog I am surprised to discover that the accuracy of ChatGPT’s search function, when coupled with its reasoning process, now exceeds that of Perplexity Pro, even when the latter utilizes all available models.
@GoogleDeepMind@lmarena_ai However, Gemini's search capabilities are very poor, far inferior to both ChatGPT and Perplexity Pro, whether in Pro or Flash mode.
@coreyyin@perplexity_ai Integrated search, but with their filters, I found the models were no longer distinct. If they provide incorrect input, the model will also be incorrect, unlike the inherent capability of the original model as demonstrated in the original application.
@perplexity_ai If these search results are incomplete or inaccurate, the models will be unable to answer correctly. For a question posed to Gemini 2.5 Pro on the Gemini platform, it gave the correct answer, whereas on Perplexity Pro, the answer was wrong.
@camerontstow Although the models are integrated, I think their own sonar is still the most accurate, combined with truly excellent reasoning. However, the reasoning model seems to be made less "smart" by the search data.