GPT Image 2.5: Two MTL Exits Intros, Before and After

You know the station. You know the line. But when the train arrives, there is still one small Montréal question: where should I stand?

That is the problem behind MTL Exits—and the story we wanted its intro video to tell. With GPT Image 2.5 arriving, we have two versions to put next to each other: the original, made before the update, and our latest version, made after it.

The most interesting difference is how quickly the imagery puts you in the rider’s shoes.

Watch the two versions

V1: the original intro

Watch the original MTL Exits intro on YouTube.

Latest: the platform story

Watch the latest MTL Exits intro on YouTube.

What changed on screen

The opening gets closer to the problem. V1 begins with a warm window-side scene overlooking Montréal: a blue book with a train symbol, the MTL Exits name, and “Ride smarter. Exit faster.” It gives the app a friendly sense of place. The latest opens with a rider wearing an orange backpack beside a blue métro train, under “Stop guessing where to stand.” You see the situation and the promise together.

The demonstration keeps its setting. In v1, the opening gives way to pale backgrounds, a numbered train diagram, and instructional panels. At the walking step, the diagram and countdown do most of the explaining. The latest keeps the platform and train behind its guidance cards, with navy, blue, and orange connecting the scene to the headings. To our eye, that makes the sequence feel more like one continuous product story.

The same practical detail survives the makeover. Both versions show the example of moving from Car 7 / Door 3 toward Car 5 / Door 2, with 56 steps and a walking countdown. V1 gives that diagram room on a quiet background; the latest adds a stronger sense of being there. There is a tradeoff: richer imagery has more to compete with the instructions. The useful test is still whether a rider can understand the next action at phone size.

Our preference is the latest opening for introducing the app to someone scrolling past. It makes the rider’s question immediate. The original’s cleaner instructional slides remain a useful reference for explaining the details.

Where GPT Image 2.5 fits

OpenAI’s official documentation describes two GPT Image 2.5 models: Sunburst, aimed at image generation and editing where precision matters, and Flare, aimed at fast, high-quality everyday image generation. Both accept text and images and produce images.

For a short app introduction, that matters at the visual-asset stage: creating a setting, refining a composition, and exploring how a product should look in a scene. Timing, motion, typography placement, and the final edit still shape the finished video.

These two intros are a before-and-after creative comparison, not a controlled model benchmark. The finished videos alone cannot tell us how much of the change came from the model, prompting, asset selection, or editing. What we can judge is the result: the latest makes the platform experience a more consistent part of the explanation.

In Montréal? Try it on your next métro trip

MTL Exits helps you choose where to board for your destination, with train car, door, transfer-position, and street-exit guidance. Core route and exit guidance works offline after installation, and the app is available for iPhone and Android.

If you’re in Montréal, check out MTL Exits. Pick your next trip and see where to stand before the train arrives. Ride smarter. Exit faster.