Is OpenAI’s GPT-5 Finally On Its Way?

OpenAI CEO Sam Altman gave the strongest clue yet that GPT-5 is close. On Sunday, Altman posted a reply on X featuring a screenshot of what looked like GPT-5 in action. It showed ChatGPT responding to a user about whether it liked the sci-fi show Pantheon and confirming that yes, it does.

Although the screenshot seemed casual, many took it as a soft launch teaser. OpenAI has kept GPT-5 tightly under wraps, but users and competitors have long been waiting to see what comes after GPT-4o. GPT-5 is expected to handle more complex tasks on its own and work across multiple media formats.

This will also be the first time GPT-5 enters public testing with clear expectations around what it can do. Early leaks suggest a longer memory window, more accurate context handling, and better performance in visual and file-related tasks. Altman’s post may have been light-hearted, but it showed OpenAI is ready to start letting GPT-5 out into the world.

OpenAI is facing some serious competition in the AI game. Companies like Google DeepMind, Meta, xAI and Anthropic have all pushed out advanced tools in recent months. GPT-5 could be OpenAI’s way to maintain its edge… at least for now.

 

What Has ChatGPT Released This Year So Far?

 

Before the GPT-5 preview, OpenAI made dozens of changes to ChatGPT. These updates have shaped how users interact with the chatbot across paid and free versions.

The most recent launch was Study Mode, announced on 29 July. This tool lets users turn ChatGPT into a tutor. It asks questions, checks understanding, and adapts to each user’s level. It also works with uploaded materials, like PDFs or images, to give direct help based on real examples.

 

 

Earlier in July, OpenAI rolled out Canva and Notion connectors for Pro users, and allowed chat search in HubSpot. Advanced Voice Mode was also improved and made available to free users. This version sounds more human, translates better, and handles long conversations more naturally.

The rollout of o3-pro in June was one of the models that really stood out. It replaced o1-pro and became the most preferred model in academic reviews. It handles long reasoning tasks and outperforms older models in science and coding.

 

How Has ChatGPT Changed Since GPT-4o?

 

ChatGPT’s biggest release before GPT-5 was GPT-4o, which replaced GPT-4 in April. GPT-4o was more conversational, better at following instructions, and could generate images. It worked across text, audio, and visual formats, making it the most flexible model OpenAI had released.

Since its launch, OpenAI has updated GPT-4o multiple times. In March, changes improved its coding abilities and made answers more concise. April brought fixes for problems with overly agreeable responses. May and June updates improved its memory use and added better formatting and clearer instructions.

GPT-4.1 followed soon after and was especially good for web development and precise tasks. Users could switch between 4o and 4.1 depending on what they needed. A mini version, GPT-4.1 mini, was also released for lighter tasks and faster replies.

 

What Comes After This Soft Launch?

 

Altman’s screenshot may be the first public peek at GPT-5, but it arrives after a busy stretch for OpenAI. From connectors to voice upgrades, the service has grown more layered and interactive. There is growing attention on how GPT-5 will handle files, memory, visuals and autonomy compared to GPT-4o and o3-pro.

So far, OpenAI has not shared a release date or technical specs. But the attention around the screenshot and OpenAI’s pattern of gradual releases hint to a rollout before the end of the year.