I’ve been using chatGPT’s API for various micro functions, and I was surprised to find how slow it was and to then learn that when called via an API it can’t browse the web.
I’ve started using Perplexity and their Sonar LLM instead which seems blindingly fast…
Do you have any other recommendations for other LLMs that are super fast over API and can also browse the web?
Hi Antony,
I have been using Perplexity Pro for a couple of months now. I run Perplexity on their portal and I have found Gemini 2.5 Pro to be much better in speed, code suggestions and improvements.
I used ChatGPT5 in Perplexity a couple of times. Although it performed similar to Gemini, the responses stopped appearing properly formatted and sometimes with missing characters after 4-5 queries in the same thread.
Nice to see you Antony. I'm not really a fan as not great for my specific tasks by you want fast then try Grok Fast 1 (free for a few more days). Its speed is truly astonishing and feels like other LLMs are running across a 300 baud modem and Grok is on full 10gb fibre!