Racing the 2025 Long Beach Grand Prix in Automobilista 2

Unless you spent your weekend living under a rock, you will know that the second week of April was the West Coast’s time to shine in the American motorsport scene. With the streets of Long Beach built up into a racing circuit, both IndyCar and the IMSA Sportscar Championship paid the city a visit for their respective third rounds.

Well, with Automobilista 2 getting an accurate version of Long Beach in its first Racin’ USA Pack almost four years ago as well as a totally not and Indycar in the Pack’s third instalment, plus a collection of IMSA-themed content filling the coffers late last year, the game is now a serious top challenger for recreating the event’s headline races.

The Long Beach Grand Prix Automobilista 2.

Always looking for more immersion and aiming to recreate the latest top motorsport events, that collection of content gave me an idea. Could I feasibly replicate the entire 2025 Long Beach Grand Prix weekend in a single evening? Well, I gave it a good go, so here is how it went.

Setting up the Weekend​

With only a few hours in my evening set aside for racing, I could have focused on a single series and completed a full-length race. But looking to experience both IndyCar and IMSA, I decided to halve both race distances and double up both fuel and tyre wear in the game’s settings.

This meant that from the standard 100-minute sprint, my IMSA race would last just 50 minutes and where a standard Indycar battle at Long Beach is 90 laps, I went for a more manageable 45.

Utilising the tyre wear and fuel usage multipliers, however, strategies would remain the same as the real-world events, meaning the IMSA race should be doable on a single stop towards the middle, whilst the Indycar race is a balance between a flat-out three-stop or heavy fuel-saving two-stopper. Choosing to match the AI, as seen in previous testing, the more aggressive option was my choice.

GTD field into Turn 1 at Long Beach.

Finally, to enjoy the race without having to think too much while narrowly avoiding concrete barriers, I decided to use just a single weather slot with bright, sunny conditions. It is the West Coast, after all.

With an hour-long practice session, ten-minute qualifying, and difficulty settings perfected—or so I thought—it was time to hit the track. Much like the real event, IMSA’s multiclass brawl came first.

An IMSA Failure​

As ever with longer-form offline races in Automobilista 2, my evening’s proceedings were not as straightforward as simply getting behind the wheel and racing. Before committing to the race, a lot of testing took place to perfect AI difficulty and strategy. However, despite managing to run a full hour-long practice and placing in the mid-GTP pack, qualifying was not as competitive.

GTP cars at the Long Beach fountain in Automobilista 2.

Whilst a nice idea, Reiza has implemented its AI’s qualifying run simulation in a way that often stumbles over itself, constantly blocking other racers on track when either on a slow out- or in-lap. As a result, I always run private qualifying to ensure times are somewhat representative. Despite this, however, my computer rivals decided to take qualifying at a leisurely pace, meaning that without pushing, pole position was an easy win for myself.

“Okay, we’re going for the Porsche Penske simulation tonight”, I muttered to myself as I decided to hit ‘Start Session’ and get the race underway. But a Porsche Penske perfect start, it was not. All of a sudden, from their dismal qualifying pace, the AI seemed to find multiple seconds on race pace from Lap 1.

That, paired with an odd starting procedure at Long Beach that both messes up the grid and causes the AI to not recognise the player’s car, driving straight into you in Turn 1, meant that the first to last challenge was indeed complete.

The AI is not the best at multiclass racing in Automobilista 2.

For the first few laps, I could just about stay with Tijmen Van Der Helm in the JDC-Miller Porsche 963 at the back of the field. But when it came time to lap the GT traffic, it all went down the drain. In Automobilista 2, the AI does not have the ability of Le Mans Ultimate or iRacing to pass slower traffic, and paired with the tight nature of Long Beach, it is fair to say this was an easy way through the pack.

That is until the second pass through traffic, when the tight final hairpin took its first victim. A dive from a car ahead on a slower class car resulted in total gridlock just into the braking zone—something I and many other GTP cars did not see in time, claiming the lives of many racing cars.

JDC Miller is always the underdog in IMSA, even in AMS2.

With a difficult start to the race, meaning any sense of immersion and fun was very much out the window, I am not shy to admit that this incident brought more joy than sadness. Finally, I could end the suffering of simply lapping for the sake of it and move onto the more fun of the two races: Single-class Indycar.

Indycar Long Beach GP in AMS2​

Much like the IMSA race, my Indycar recreation started with an hour-long practice session to dial in a setup and perfect my strategy plan. Always struggling with traction versus the AI in Formula USA 2023, that setup meant an extremely soft rear end and locked differential, helping me put the power down, notably out of the final corner.

In fact, I struggle so much with traction in the fake Dallara IR-18 that where I can run most car classes around the 110 per cent AI difficulty mark, I run these at 98.

A Slow-Burning First Two Stints​

However, once I felt comfortable with the car, it was time for a quick qualifying session, where my fastest lap placed me in P23. Once again, the quirky rolling start procedure at Long Beach meant that by Turn 1, despite some AI argy-bargy, I would be sitting in P16 with Conor Daly up ahead and Pato O’Ward close behind.

Racing Pato O’Ward and Conor Daly on the streets of Long Beach – Only in AMS2.

With O’Ward on my tail for most of the first two stints, I spent the first half of the race very much with my eyes on the rear-view mirror, meaning Daly ahead was slowly opening a gap. Not quite maximising the 15-lap tank of fuel, the entire field pitted around lap 13 for the first stop, while I managed to go one lap further.

All-Action Second Half​

It was with about 20 laps to go that the second cycle of pit stops took place, and looking to jump my closest rivals, I chose to take just 10 laps’ worth of fuel, making for a shorter stop – essentially splitting the last half of the race in two. And it worked as exiting the pits, Daly was just hitting the brakes into T1, though this meant I had colder tyres and sadly could not keep the position. Eventually, the McLaren also got past, leaving me to make on-track overtakes if I wanted a top-15 position.

And versus the Arrow McLaren, it seemed that I was fast through Sector 2, getting onto his gearbox down Seaside Way multiple laps in a row. Slower on the straights, however, I knew it would have to be a lunge down the inside until I got closer than ever – and even alongside. Though now, it was time for another oddity in AMS2 AI behaviour as despite being alongside for much of the braking zone and even ahead at the apex, Pato turned in as if I was not there, sending me into a half-spin.

The AI in Automobilista 2 squeezing a bit too much at Long Beach.

This is something I have already pointed out as a flaw in AMS2 AI racecraft in an AI comparison between AMS2, LMU and iRacing. The AI tends to act as if you are not on track, taking its normal line unless you are far ahead into a corner. Banging wheels, the Madness Engine then comes into view, somehow cancelling out any front-end grip, meaning you barrel straight on.

After my half-spin, I found myself a good five seconds behind my previous rivals and lost a place to Santino Ferrucci. Luckily, the 17th-place man appeared to lack McLaren’s straight-line speed, so a move into Turn 1 was swift.

Next, another Automobilista 2 flaw became rather prominent as lapped traffic and flashing blue panels appeared. Since its most recent update, AI cars that find themselves a lap down are supposed to move to one side of the track on the straights and slow down. But it seems they only do this just before a braking zone, and so you often find yourself going side-by-side with slower traffic, at which point they speed up again. In addition, this behaviour seems to confuse faster cars, and so you end up with large groups of cars tripping over each other. A nice bit of immersion, this is something that I feel we can live without as, alongside the in- and out-lap behaviour in practice and qualifying, it tends to ruin results.

Chasing cars ahead is a lot of fun.

So with my previous competitors stuck in traffic and lacking commitment, I could easily make my way through, eventually getting into P11 with three-or-so seconds to Felix Rosenqvist. Five laps to go, it was challenge on.

But sadly, I did not quite have the pace to catch and pass. I only really closed up on the last lap and was unable to make a move before the chequered flag.

AMS2 AI Conclusion: Fun but Flawed​

As is usually the case with these more focused offline AI races, I have come away from my time at Long Beach with a sense of dissatisfaction and a long list of questions and concerns.

Obviously, I have learnt by now that perfect AI in any game is pretty much impossible. You will always come across oddities and poor behaviour – just as you would in online racing. But it seems that an otherwise perfect formula has been hindered by a few overwhelming issues that have really grated on me.

Slow AI is a disappointing issue in Automobilista 2.

These mostly centre around the game’s desire to simulate real-world racing etiquette, from overly ambitious blue flag scenarios to flying lap versus cooldown overlaps. Be it in qualifying or the race, results can get easily skewed by slower cars getting in the way by trying to get out of the way. Ironic, right? Without these fancy, over-the-top additions, both single-class and multiclass racing would be so much better, as even the GT traffic against Hypercars, for some reason, decides to slow down in strange places, stacking up the field.

But it isn’t just blue flags that mess up the AI experience in Automobilista 2. While I can deal with AI difficulty strength varying from one cat-track combination to another, the sudden change in speed between sessions in the same race weekend really grinds my gears. If I qualify in the top 15, I expect to be racing with those cars in the top 15, not to either drop to the back and struggle all race or jump to the front and build a ridiculous gap.

It may be flawed, but racing the AI in Automobilista 2 can be excellent fun in the right scenario.

It is far from all bad, however, as, despite these immersion-breaking issues, the overall experience was a lot of fun, especially in the IndyCar event. The thrill of pushing your car to the limit in hopes of catching a rival ahead or maximising your push to pass just to stay ahead of a name you recognise from the real world is immense.

You may remember an alternative 2025 Indycar calendar we posted last week, showing how to get a full season using the Formula USA 2023 car in Automobilista 2. Following this Long Beach extravaganza, I may well try to follow the season throughout the year, tracking my progress.

Do you enjoy recreating real-life races in sim racing? What is your go-to series and sim in which to replicate it? Let us know in the comments.