Praga R1 Indianapolis 2012 Grand Prix 11 Adaptive AI (AI level probably 96) 15min race standing start Sadly no screenshot of results. 1) Although the Praga has a fixed gearbox, AI was faster than me on the straight. Full throttle and in the limiter, they were just faster. 2) AI was very slow through the chicanes.
It's missing for the Praga, but I should have them for at least some of the DTM stuff. I'll add them.
STW Zandvoort(Modernized) 27 AIs 97% difficulty Standing start 6 lap distance with a 12 minute long qualifying AIs are superior on this track through corner 1 Tarzanbocht and slightly at Hugenholzbocht Also on the approach towards Tarzanbocht, with cars that lack ABS they are much better on the brakes than the player Qualifying results Race results
can you say if the distance between yours and the ai times is more exagerated in the first 3/ 4 laps Andii
Ai% 93 how many ai - 31 AI cars car and track GT4 at Monza GP distance 40min rolling/standing start - rolling AI BOP -. There were 8 Porsche caymen gt4 cs mr and the 7 of them finished 1 thru 7 other cars were equally split between Audi,BMW, AMG, McLaren
At this moment, Salzburgring is really good with racing at a skill level between 95-97 and am so loving it! But sadly, during qualifying you still can run into an AI bugging out! Class - STW Number of AI - 29 Length - 30 minutes At this moment, the AI O.Nakako just got some insane pace put an insane lap time over the rest of the AIs. But during the race, the same driver slipped down to fourth. Also worthy mentioning is, this bug also happens on this track even during races but only if the difficulty is atleast 98! Will show tomorrow what it can look like since I am writing this before bedtime! So my sincere apologies
NEW CAR : NEW ISSUES Brands Hatch, 98%, ALPINE A110 CUP, 29 AI, standing start AI CRASH every lap at the turn 6 (they seem to struggle with pack race and this turn, and to be honest I have seen this issue so much times with this track or anothers with different cars since 5 years)
So after the update, the AI race behavior has basically become a lottery. On some tracks, like Oschersleben, the AI is much slower in races not only compared to qualifying, but even compared to practice sessions. On other tracks (for example Adria 2003), the AI in races runs almost at qualifying pace. (it was always like that before the update and it was okay) Running a championship against the AI in these conditions is impossible, honestly. It feels like my favorite game got messed up — hopefully they fix it soon and are aware of the issue. (I tested this with different classes, with and without tire/fuel consumption, in long or short races)
Some more information about the AI pace issue in races after the Q1 2026 update The lap times reported below are based on the average of the 3 fastest AI cars... set with the five Lamborghini of DTM 2024, with fuel and tyre wear x1 and after two full laps completed (however, the issue is exactly the same with WTCC cars, hypercars, or the entire DTM roster — so this is a general problem, not tied to a specific car). On some circuits, the AI behaves as before I mentioned Adria 2003, but for example it happens also with Norisring or Nürburgring Sprint. Here, with AI level 92, the top three AI drivers lap in around 1:29.3 in qualifying and then maintain the same pace in race conditions (15-minute sprint race). This is how the game worked before the update — the AI was basically as fast in race as in qualifying, which made sense given the limited fuel differences. On other tracks, everything changes (Oschersleben Alternate or Bilster Berg Gesamtstrecke) For example, at Bilster Berg Gesamtstrecke (no chicane, AI level 92): 1:39.2 in qualifying (average of top 3) 1:41.2 in race That’s a 2-second drop, making it very easy to beat the AI by recovering positions compared to qualifying. Other example, LM Hypercars full grid in Oschersleben Alternate 2024 top 3: 1:20.4 in practice / 1:18.5 in qualifying / 1:21.4 in race BUT, IMPORTANT! → If you increase the AI level, the issue disappears. Bilster Berg Gesamtstrecke DTM 2024 Lamborghini: Level 100 top 3: 1:36.2 quali / 1:36.2 race Level 97 top 3: 1:37.1 quali / 1:37.2 race At level 95, it becomes a middle ground: 1:37.4 quali / 1:38.3 race So overall, there is a huge inconsistency, whereas before everything was consistent across all tracks. This difference based purely on AI level did not exist before. Someone might say “just increase the difficulty,” but for players like me using a gamepad (and no setup changes), it’s not easy to run at very high AI levels on every circuit. Right now, this creates a strange situation that makes it almost impossible to run a proper offline championship against the AI. I just hope the developers are aware of the issue and can address it.
Class: STW Number of opponents: 27 Skill: 95 Track: AVUS(1998) Length: 12 laps Start type: Standing This issue with AVUS has existed since launch. Although I did report about it before, I just wasn't specific about what happened since I didnt look through in detail. But here it is again: The AIs have better handling and are also able to accelerate way faster out of the last corner Nord Kehre. As you can see in more detail in this video, the AI can just floor it and glide on as if it's got a boost or something. The player can never even gain remotely that speed the AI can do whilest turning! Same also happens at the 1994 version too. @Alex Hodgkinson the best thing that can be done to solve this is to force the AI to lift throughout the corner to try solving this! But I hope this video explains the issue well. Also don't forget that, on turn 11(Kink) at Road America and at Mutkurve on Nordschleife I have seen AIs doing the exact same thing, that the AI driver can gain speed whilest turning!
Good spot, this was caused by AI over estimating how much fuel was needed and running very heavy in races. This is due to virtual energy requiring low engine maps in races
Is that just for the ai because they cant come up with better strategies or even for the player? Because as a player i feel like its almost always better to just run at full pace and do a pitstop (more). The only scenario where saving VE seems beneficial would be if the race is like 1-2 laps longer than going at full pace would allow you to and even then mainly on tracks with a long/slow pitlane. Getting back to the ai, i did one such race against the ai earlier this year (40 minutes at daytona, see middle of page 1 of this thread) and there the ai actually pitted while saving some VE and not pitting wouldve been faster
No it's how the real Hypercar class works; virtual energy is the limiter for lap times and it's very restrictive. Cars can't run faster and do extra pit stops in the real world because total energy use is capped and our AI run at about that capped pace.
Ive never heard about total energy being capped (total as in the maximum allowed for the whole race) and also couldnt find anything about it now. Obviously theres a per stint energy limit but thats where going faster and pitting more frequently comes in. Either way id think itd make sense for the in game ai to use the best in game strategy. Im aware drivers tend to save energy quite a bit in real life but in real life you have a bunch of factors that are non factors in raceroom (safety cars and corresponding pit closures, weather, limited tire sets, driver times, usually longer races etc). In game there basically isnt anything unexpected (and no limit on total energy from what i can tell) so you can just optimise your total race time whereas in real life you gotta be flexible and prepared for a wider range of scenarios. And combine that with the possibility of any slight deviations in simulation that means a different strategy is ideal in raceroom than in real life. PS: we still have no option to replenish VE without taking on extra fuel. Hope something can be added there PPS: if this is getting too off topic we can also move it here https://forum.kw-studios.com/index.php?threads/about-virtual-energy.20749/#post-257132