Openai's Chatgpt plans for GPT-5: Integration of tools and update of the operator agent

Published on: May 20, 2025 / Update from: May 20, 2025 - Author: Konrad Wolfenstein

Openai's Chatgpt plans for GPT-5: Integration of tools and update of the operator agent-Image: Xpert.digital

Openai GPT-5-a platform concept for the next generation

The future of the AI: what GPT-5 from Openaai will change

Openai is developing GPT-5 as the next large voice model that is intended to improve existing functions and combine various AI tools in a uniform platform. The focus is on the seamless integration of Codex, Operator and Memory-with the aim of offering users a simpler and more efficient AI experience.

Suitable for:

GPT-5: The next generation of artificial intelligence of Openai-progress in thinking and performance

The vision of a standardized AI system

Jerry Tworkek, Vice President of Research at Openai, describes GPT-5 as the next “Foundational Model”. It should make "everything our models can currently do better - and with fewer model changes". This statement indicates a pragmatic approach in which the focus is on improving the user experience.

The original vision of Openaai was ambitious: GPT-5 should fully merge the previously separate GPT and O model series. However, this strategy had to be adjusted. In April 2025, Openai realized that the integration of all components was more complex than expected, which led to a shift in the GPT-5 releases and the separate publication of the Reasoning models O3 and O4-Mini.

The current goal is to link the different tools in such a way that they feel like a uniform system for users. TWOWEK explains: "The aim is to reduce the manual use of various models and tools that have been necessary so far". Instead of complete technical integration, GPT-5 should at least offer at least one standardized user experience.

Openai officially announced the shift of GPT-5 on April 4, 2025. CEO Sam Altman confirmed this in a post on the platform X :

“Change of plan: We will publish O3 and O4-Mini, probably in a few weeks, and then GPT-5 in a few months. There are several reasons for it, but the most exciting thing is that we can do GPT-5 much better than we originally thought. We also found that it was more difficult to integrate everything than expected. And we would like to make sure that we would have to have enough capacity to cover unprecedented demand. ”

Important components of integration

The integration includes several key components:

Codex: The new code model should work closely with other tools. Codex has already appeared as an independent tool and offers various safety modes such as “Suggest”, “Auto Edit” and “Full Auto”.
Deep Research: This research component is also integrated into the GPT-5 platform.
Memory system: This function is intended to store and call up personal information and previous interactions.

The operator agent: desktop assistant with new functions

The plans for the operator agent are particularly exciting. This screen agent is currently in the research phase and can already take on basic tasks such as browser control-but is not yet reliable.

Jerry TWOWEK has announced an update that should appear “soon” and could make operator a “very useful tool”. The long-term vision for operator seems to be that of a comprehensive desktop assistant: "In the long term, operator could become a switching center for daily tasks-not in the browser, but directly in the operating system".

Suitable for:

Difference between AI agents and AI assistants: A comprehensive analysis

Why the integration of GPT-5 is more complex than expected

The development of GPT-5 has several challenges:

Technical complexity: The complete integration of all components has proven to be more difficult than expected.
Increasing resource requirements: token consumption increases continuously. Openaai therefore invests massively in computing capacity in order to meet the expected “unprecedented demand”.
Balancing of costs and benefits: TWOWEK is convinced that tokens are not only more useful over time, but also cheaper.

According to current information, GPT-5 should appear “in a few months”. Publication in spring or summer 2025 seems realistic if no further technical hurdles appear.

Openai has distanced itself from classic benchmarks and instead focuses on practical tasks to assess its models. TWOWEK emphasizes: "The classic performance key figures are hardly meaningful these days. Instead, specific tasks should be used to really evaluate the models".

Pragmatic perfection? Openai and the vision behind GPT-5

Openais plans for GPT-5 reflect a pragmatic approach: The complete technical integration of all components remains a goal, but was initially replaced by a closer link between the existing tools. The focus is on an improved user experience with fewer manual model changes. The planned update for the operator agent could be a significant step towards a powerful desktop assistant who can act beyond the borders of the browser.

The coming months will show whether Openaai can implement its ambitious plans and whether GPT-5 will actually become the promised quantum leap towards a “magical standardized intelligence”.

Suitable for: