If I had to make an educated guess on some F5-EF5 tornadoes I feel reached 350+ wind speeds, I’d say Jarrell, Parkersburg, El Reno 2011, BC-Moore 1999, Smithville, Tri State, New Richmond, and Sherman are all some tornadoes I wouldn’t be surprised but 350+. That’s just my educated guess though.
I've been thinking this question for a while as well which is what's the maximum potential intensity tornado can reach on earth and what factor determine it.
There was very few valuable research about this topic but the more I learn the more I feel this paper in 2017 should be taken more seriously
I know it's just a simulation but one may need to give it a read with the comparison of this paper about hurricanes
Basically they used the same method so they were very comparable and we have much better observation for hurricanes than tornados. The Daniel P. Stern and George H. Bryan paper in 2018 seemed to produce unreasonablely large gusts on surface. But after further analyzing, one would see it's actually in quite good agreement with the observations we have. This is the reason why I said we need to take David S. Nolan's paper in 2017 more seriously even it's just a simulation.
Then, in this paper, we will find that the SR005 simulation with lower swirl ratio and smaller core of tornado than control simulation presented highest winds.
This feature alone is in good agreement with what we see in reality which further enhance my confidence in it cause most EF5 tornados we've seen in reality tend to have a small core.(vertical components in this type of tornados are vastly stronger than tornados with larger swirl ratio).
The simulation showed mean 3s horizontal winds well over 130m/s and mean 3s vertical winds well over 120m/s in a 5min span. When the vortex being the strongest, it even had over 220m/s instantaneous horizontal winds and over 225m/s vertical winds at the same time.
Also you will find in this paper that with friction increased, the peak winds only slightly increased compared with no friction simulation which also well accord with other researches.
The one biggest question here is could any tornado in history once had azimuthal-mean wind of 104.9m/s which is the initial input of SR005 simulation? In Mulhall tornado, we see 84m/s azimuthal-mean wind at 0310z and in Spencer there was similar results which are already pretty extreme but we know that it's highly unlikely that Mulhall or Spencer were the strongest tornado ever. Still, one couldnt say for certain that there ever exist a tornado with that strong azimuthal-mean wind.
The second problem is what role will debris play into this circumstances. There were
contradictory results about its impact on tornado wind speed, mainly between the Lewellen 2008 paper and Shawness tornado 2020 paper. But one thing for certain is debris can only enhanced the damage potential or the overall momentum regardless of wind speed.
Finally, why DOW never detected winds that extreme like in simulations? One biggest point is there is almost no vertical component winds on radar when it's scanning at lower angle. The second one is they don't get true ground wind speed at 99.9% time. The third point is what DOW detected is the reflectivity-weighted mean velocity in certain volume rather actual air flow.
Even with all these papers and researchs, one should still know that we are far from truly know about tornados' actual winds and either speculation can be correct.
