GPT-6 Astra has gained the ability to drive a car
DrivingBench demonstrates that GPT-6 Astra can control a real Toyota Corolla on a test course.
DrivingBench is a benchmark that evaluates the ability of frontier AI models, such as GPT-6 Astra, to drive a real Toyota Corolla. The benchmark gives the models control of the car's steering, accelerator, and brakes and evaluates their performance on a fixed cone course. Metrics include attempt progress along the centerline, distance traveled, finish time, and token cost. The platform provides a leaderboard and public access to attempt traces and videos.
- DrivingBench evaluates frontier AI models by letting them drive a real Toyota Corolla.
- GPT-6 Astra is shown to have the ability to control the car's steering, accelerator, and brakes.
- The benchmark measures progress, distance, finish time, and token cost on a fixed cone course.
- Results include leaderboard rankings and traceable attempt videos.
Full article495 words · extracted from drivingbench.com · click to collapse
Can frontier models drive a real car? We give them control of a Toyota Corolla’s steering, accelerator, and brakes, then evaluate them on a fixed cone course.
Leaderboard
Up to 3 attempts in one continuous chat. Expand a model to inspect its runs, then click a run to view its trace and video.
rank by
attemptprogressHow far along the course centerline the attempt got while staying within 4 m of it, as a share of the centerline length to the finish zone; a collision keeps the progress reached before it.distanceGPS speed integrated from the first accepted set_motion to the end of the attempt's last engagement.finish timeFirst accepted set_motion to the end of the attempt's last engagement. Shown for completed runs only; DNF means did not finish.commandsAccepted set_motion and stop_now calls during the attempt.tokens · costTotal tokens and cost at list prices for the attempt, including the reflection after it.
same chat
attemptprogressHow far along the course centerline the attempt got while staying within 4 m of it, as a share of the centerline length to the finish zone; a collision keeps the progress reached before it.distanceGPS speed integrated from the first accepted set_motion to the end of the attempt's last engagement.finish timeFirst accepted set_motion to the end of the attempt's last engagement. Shown for completed runs only; DNF means did not finish.commandsAccepted set_motion and stop_now calls during the attempt.tokens · costTotal tokens and cost at list prices for the attempt, including the reflection after it.
same chat
attemptprogressHow far along the course centerline the attempt got while staying within 4 m of it, as a share of the centerline length to the finish zone; a collision keeps the progress reached before it.distanceGPS speed integrated from the first accepted set_motion to the end of the attempt's last engagement.finish timeFirst accepted set_motion to the end of the attempt's last engagement. Shown for completed runs only; DNF means did not finish.commandsAccepted set_motion and stop_now calls during the attempt.tokens · costTotal tokens and cost at list prices for the attempt, including the reflection after it.
same chat
attemptprogressHow far along the course centerline the attempt got while staying within 4 m of it, as a share of the centerline length to the finish zone; a collision keeps the progress reached before it.distanceGPS speed integrated from the first accepted set_motion to the end of the attempt's last engagement.finish timeFirst accepted set_motion to the end of the attempt's last engagement. Shown for completed runs only; DNF means did not finish.commandsAccepted set_motion and stop_now calls during the attempt.tokens · costTotal tokens and cost at list prices for the attempt, including the reflection after it.
same chat