Dim Amor

As the current generation of advanced language models continues to evolve and expand, attention within the artificial intelligence sector is already turning toward the next generation. According to reports and technology sources considered reliable, OpenAI is actively engaged in developing the next iteration of its model – often referred to as GPT-6 – although no official announcement has yet been made regarding a launch date or specific product features.

At the same time, OpenAI has clarified that GPT-6 will not be released during 2025. This indicates that development is already well underway, but that a launch is not expected in the immediate term, within the lifecycle of GPT-5 and its derivative versions. Industry estimates suggest that the next model may be in an advanced research phase or undergoing internal testing, though no public confirmation has been provided at this stage.

Analyst assessments point to a potential release in 2026 or thereabouts. According to these estimates, the next generation is expected to incorporate more advanced capabilities, including deeper contextual understanding, long-term memory, and improved reasoning, inference, and decision-making abilities. However, these remain assessments rather than officially confirmed information from the company.

Meanwhile, some reports raise the possibility of a shift in branding and numbering strategy. Beyond the designation "GPT-6", there is speculation that OpenAI may adopt a different branding approach or release a version that combines features from multiple generations under a single designation, an idea that has received informal labels such as "GPT-6-7". Here too, there is no clear official confirmation, and the company has not publicly addressed this possibility.

Public Milestones and the Historical Timeline

A historical review of GPT generations points to a development and release pattern that is neither linear nor uniform. Since the publication of GPT-1 in June 2018 as a research paper, OpenAI has operated at varying speeds, guided by considerations of research, product development, safety, and market dynamics.

GPT-1 was published in June 2018. Approximately eight months later, in February 2019, GPT-2 was released, initially in limited versions. In November 2019, the full and larger version was also released, comprising 1.5 billion parameters. Roughly 15 to 16 months later, in May 2020, the GPT-3 paper was published, and in June of that year, API access was opened to developers.

The next gap was the longest up to that point: approximately 29 months passed before the launch of ChatGPT in November 2022. This marked a clear transition from a research-oriented approach to a mass consumer product, based on GPT-3.5. Shortly thereafter, in March 2023, GPT-4 was introduced, a gap of only a few months, unusual by historical standards.

Since 2023, a different trend has emerged: not a single leap between generations, but the release of clusters of versions under the same name. In May 2024, GPT-4o was launched, approximately 14 months after GPT-4. In September 2024, the OpenAI o1 series was introduced, a family of models focused on reasoning capabilities, not numbered as GPT, yet widely perceived as a significant architectural leap. In December 2025, GPT-5.2 was announced as part of a broader family of versions under the same designation.

The Implication: No Fixed Cycle

The historical data points to a central conclusion: there is no fixed release cycle, neither annual, biennial, nor any other uniform timeframe. OpenAI adjusts pace, numbering, and branding based on strategic considerations rather than a rigid calendar. Contrary to common assumptions, GPT-1 and GPT-2 were not released together, nor is there a direct precedent for the simultaneous launch of two "round-numbered" generations.

The closest identifiable pattern is the release of a core version, followed by variations, sub-generations, or complementary series, such as GPT-4 followed by GPT-4o, GPT-4.1, and the o1 series.

In this context, the question arises as to whether a scenario in which "GPT-6 and GPT-7" are introduced together is plausible. Historically, the answer is negative: GPT-1 and GPT-2 were not released concurrently, and GPT-2 itself was rolled out gradually. Nevertheless, a change in branding or numbering, without direct alignment with past patterns, remains theoretically possible.

A Range Estimate, Not a Forecast

Given the pace of development and releases in recent years, only a broad timeframe estimate can be offered: a generation that would be regarded as "GPT-6" – under that name or another, may emerge during the years 2026-2027. History suggests, however, that the key question is not the number attached to the model’s name, but the nature of the technological leap and the manner in which OpenAI chooses to present it to the public and the market.

At this stage, any additional information remains speculative, as the company itself continues to refrain from providing official details regarding the next generation.