I wouldn’t go as far as saying lies more like inaccurate which likely describes the results better. In fact when using ookla from the web browser - the calculations is always showing a better speed than my subscription allows for - which makes me think the algorithm for the ookla web test is too generous - but I have no way of knowing for sure. I have noted the differences with speedtest-cli as well - but it is nothing to worry about - it likely due to differences in calculations and you should not put too much trust into them. This is just the scenarios I could pull from memory - there’s likely more.Įven if you are using a browser based test - the browser may influence the test - e.g.browser plugins/add-on. is the test executed connecting directly to the ISP (without local routing).is the test executed through a local router.the method used to calculate the bandwidth.any given activity on the local network.ISP load balance or endpoint load balance. the actual route - which may change between tests - e.g.It is very difficult to identical results as any bandwidth test will be influenced by When calculating the speed of a connection - there’s a lot of factors. It can’t be a coincidence, I tried several times at different timesĪlso I tried “no-install” version of cli: $ curl -s | python. Testing from Turkcell Superonline (**********). I tried rebooting the router, computer and replugging the wifi adapter, but the results are equally bad in the cli versionĬli results: Retrieving configuration. I noticed that speedtest-cli-2.1.3-2 and produce completely different results.

Speed test upload ookla