Formula 1 pre-season testing for 2026 concluded in Bahrain with another smattering of race simulations, with virtually every team having attempted some clear grand prix-like running at some point during the six days in Sakhir.
Programmes differed hugely across the board, as they do, and some race simulation plans either went unrealised or were interrupted.
What data we do have is plentiful but, as always in testing, extremely caveated.
Some ran in the morning, some in the afternoon; some two hours earlier, some two hours later; some two-stopped, some three-stopped, some caught a red flag; and some didn't hit anywhere near the grand prix lap count, making it fair to wonder what fuel level they really started at - which is before you even think of engine modes, energy delivery strategies and the likes of.
However, here's what we have from the final day - and how it fits into the overall picture from the test.
Race runs (or similar)
Lando Norris, McLaren
Start: 6.04pm
Stint 1: C3, 10 laps (1m39.079s avg)
Stint 2: C2, 15 laps (1m38.344s avg)
Gabriel Bortoleto, Audi
Start: 5.13pm
Stint 1: C3, 12 laps (1m40.790s avg)
Stint 2: C2, 23 laps (1m39.987s avg)
Stint 3: C1, 10 laps (1m37.909s avg)
Stint 4: C3, 5 laps (1m36.908s avg)
Ollie Bearman, Haas
Start: 3.28pm
Stint 1: C3, 14 laps (1m41.239s avg)
Stint 2: C1, 22 laps (1m40.779s avg)
Stint 3: C2, 19 laps (1m38.987s avg)
Arvid Lindblad, Racing Bulls
Start: 3.24pm
Stint 1: C3, 12 laps (1m41.219s avg)
Stint 2: C2, 21 laps (1m41.285s avg)
Stint 3: C2, 19 laps (1m39.710s avg)
Carlos Sainz, Williams
Start: 3.24pm
Stint 1: C3, 13 laps (1m42.335s avg)
Stint 2: C2, 14 laps (1m41.375s avg)
Sergio Perez, Cadillac
Start: 12.23pm
Stint 1: C3, 14 laps (1m43.264s avg)
Stint 2: C2, 17 laps (1m42.439s avg)
Stint 3: C1, 15 laps (1m41.690s avg)*
Esteban Ocon, Haas
Start: 12.21pm
Stint 1: C3, 12 laps (1m41.111s avg)
Stint 2: C2, 22 laps (1m40.292s avg)
Stint 3: C1, 21 laps (1m40.098s avg)*
Charles Leclerc, Ferrari
Start: 11.20am
Stint 1: C3, 10 laps (1m39.039s avg)
Stint 2: C2, 13 laps (1m38.682s avg)
Stint 3: C3, 12 laps (1m37.757s avg)**
Stint 4: C2, 13 laps (1m37.139s avg)
- outlier lap deleted
** stint interrupted by Antonelli red flag (then continued)

Norris's two-stint run is fairly impressive, but has to be taken with the caveat that McLaren will have known he won't have enough time left in the session to do the full race - so didn't have to fuel his car up with the full race load, but still could have.
The waters are muddied for Leclerc's run, too, despite a greater sample size, as he ran at a different time to most and, like team-mate Lewis Hamilton the day before, completed four stints. Like Hamilton, he too had his simulation interrupted by a red flag.
Accounting for that, his run is well within the margin of error from all those factors when compared to what Oscar Piastri and Max Verstappen put together the day prior. It is entirely plausible that those teams are all close, but also equally plausible that the morning conditions undersold the potency of Leclerc's run.
The midfield teams are clearly cut adrift, but have laid down enough markers to suggest a good fight of their own.
Haas has been the consensus pick as F1's fifth-best this week, and Esteban Ocon and Ollie Bearman both completed final-day race runs (fairly analogous ones, too), which compared favourably to Arvid Lindblad's in the Racing Bulls but surprisingly not very favourably to Gabriel Bortoleto's in the Audi.
Bortoleto might have run at a more advantageous time, but the Audi was also notably quick in race trim in Nico Hulkenberg's hands the day before. It apparently doesn't inspire much confidence from trackside, but it could be a sneaky-good car.
There is no race sim from Alpine (though presumably it's right there in the Audi-Haas ballpark) and no full one from Williams, though the two consecutive stints Carlos Sainz did do were uninspiring to say the least.
Cadillac is adrift but got the race distances in. And while Aston Martin should be marginally quicker than that, its bigger worry right now is actually completing a race distance rather than how fast it can cover one.
To put the data from Thursday and Friday into context, here's a 'simulated' 52-lap race between every driver who did more than consecutive two race stints - with no start lap, inlap and outlap.
Many of these did not reach 52 laps, so their race time will be filled out with the equivalent number of laps based on their total average. This is not necessarily right - the further into a race you go, the quicker you are due to fuel load, tyres dependant - but it seems like the simplest way to do it.
For drivers who did four stints, an extra 23 seconds will be added as hypothetical time lost in pits.
1 Piastri (McLaren)
2 Leclerc (Ferrari) +3.170s
3 Verstappen (Red Bull) +8.849s
4 Hamilton (Ferrari) +16.029s
5 Bortoleto (Audi) +1m11.166s
6 Bearman (Haas) +1m36.857s
7 Ocon (Haas) +1m39.878s
8 Lawson (Racing Bulls) +1m44.705s
9 Lindblad (Racing Bulls) +1m54.678s
10 Bottas (Cadillac) +3m20.403s
11 Perez (Cadillac) +3m25.780s
Cutting off the 'race' at the 23-lap mark, at which Alonso stopped, accounts for the two-stint cars as follows:
Norris 1st of 15
Hulkenberg 8th
Sainz 12th
Alonso 14th
The Mercedes elephant in the room

The Mercedes W17 was the race simulation standout of week one, but favoured a different kind of long-running across these days - usually beginning with a push lap on the set, then either a double-cool or a return to the pits, then a long run.
On Friday, there was no representative Mercedes run of 10 laps or longer. But the team flexed its long-run muscles to some extent, logging the following sequences of between nine and seven laps across the day.
Antonelli - 1m36.856s avg
Antonelli - 1m37.032s avg
Antonelli - 1m37.363s avg
Russell - 1m38.316s avg
Russell - 1m37.785s avg
Russell - 1m37.348s avg
Russell - 1m36.667s avg
Russell - 1m36.171s avg
That last Russell race run is a clear standout that compares favourably to any end-of-race effort by any team in any of the race simulations.
But we will have to wait until Melbourne to see what the Mercedes really looks like over a race distance.
