What OpenAI’s own documentation currently says

OpenAI’s GPT-5 API model page and API Platform Documentation both show “Latest: GPT-5.4” — not GPT-5.5.

The same official API documentation describes GPT-5 mini as a faster, more cost-efficient version of GPT-5, and recommends that most new low-latency, high-volume workloads start with GPT-5.4 mini. The GPT-5 nano documentation similarly recommends GPT-5.4 nano for speed- and cost-sensitive workloads.

On the ChatGPT side, OpenAI’s Help Center has a page titled “GPT-5.3 and GPT-5.4 in ChatGPT.” It says that when users select Instant, ChatGPT can automatically decide whether to use GPT-5.3 Instant or GPT-5.4 Thinking.

That supports a clear but limited conclusion: in the official public materials provided, OpenAI’s visible model line is still GPT-5.3/GPT-5.4. Those sources do not confirm GPT-5.5 or Spud as a public model.

So where does the Spud rumor come from?

Spud is not coming from nowhere, but the type of evidence matters.

Digit reported that OpenAI’s next major AI model is code-named Spud and may ship as ChatGPT 5.5, but the article uses tentative wording such as “may” and “likely.”

The Decoder reported, citing The Information and an internal memo, that OpenAI had reportedly finished pretraining a new model code-named Spud. That is still a reported internal milestone, not the same thing as OpenAI publicly launching a model.

Another article focused on GPT-5.5/Spud explicitly labels its framing as speculation and says there is no official GPT-5.5 release date, model card, or API pricing yet.

Taken together, the third-party sources show that there is a Spud/GPT-5.5 rumor. They do not prove that Spud has been released, that Spud will definitely be called GPT-5.5, or that GPT-5.5 would be a newly pretrained foundation model.

Why the version number does not settle the question

The phrase “new foundation model” usually points to the underlying pretrained model family — not just a product label, routing choice, or incremental model update. That distinction is important here.

A label like GPT-5.5 would not, by itself, prove a new foundation model. OpenAI’s public naming already includes GPT-5.3 and GPT-5.4 within the GPT-5 line, so if GPT-5.5 appears later, the cautious first read would be that it is another GPT-5-series step.

But product names are not technical proof. To make a stronger claim that a model is a new foundation model, outside observers would normally need official evidence such as a model page, API model ID, model card, pricing, release notes, or a formal launch post. At least one current Spud/GPT-5.5 source explicitly says those official release signals have not been announced.

In other words: even if reports of Spud completing pretraining are accurate, that does not automatically establish that OpenAI has publicly released GPT-5.5, or that GPT-5.5 has been confirmed as a new foundation model.

Evidence check

What developers and product teams should do

If you are planning model routing, cost estimates, migration work, or a product roadmap, do not treat GPT-5.5 Spud as a confirmed dependency yet. The safer source of truth is still OpenAI’s official API and product documentation, which in the supplied materials points to GPT-5.4 rather than GPT-5.5.

For low-latency, high-volume workloads, OpenAI’s GPT-5 mini documentation recommends starting with GPT-5.4 mini. For workloads where speed and cost are the main constraints, the GPT-5 nano documentation recommends GPT-5.4 nano.

Spud is worth watching, but it should not be used yet as a production promise, pricing assumption, or capability assumption. The signals that would change the picture are straightforward: an official OpenAI model ID, model card, API pricing page, release note, or product announcement. The current rumor-focused sources do not provide that full official evidence.

Bottom line

The most accurate answer today is: we cannot confirm that GPT-5.5 Spud is a new foundation model — and we cannot even confirm from the supplied official materials that GPT-5.5 Spud has been formally announced by OpenAI.

If OpenAI later adopts the GPT-5.5 name, the existing GPT-5.3/GPT-5.4 naming pattern suggests it should first be understood as a GPT-5-series progression. Whether it is also a new foundation model is a separate technical question that should wait for official documentation, not version-number guesswork or third-party rumor.

What OpenAI’s own documentation currently says

OpenAI’s GPT-5 API model page and API Platform Documentation both show “Latest: GPT-5.4” — not GPT-5.5.

The same official API documentation describes GPT-5 mini as a faster, more cost-efficient version of GPT-5, and recommends that most new low-latency, high-volume workloads start with GPT-5.4 mini. The GPT-5 nano documentation similarly recommends GPT-5.4 nano for speed- and cost-sensitive workloads.

On the ChatGPT side, OpenAI’s Help Center has a page titled “GPT-5.3 and GPT-5.4 in ChatGPT.” It says that when users select Instant, ChatGPT can automatically decide whether to use GPT-5.3 Instant or GPT-5.4 Thinking.

That supports a clear but limited conclusion: in the official public materials provided, OpenAI’s visible model line is still GPT-5.3/GPT-5.4. Those sources do not confirm GPT-5.5 or Spud as a public model.

So where does the Spud rumor come from?

Spud is not coming from nowhere, but the type of evidence matters.

Digit reported that OpenAI’s next major AI model is code-named Spud and may ship as ChatGPT 5.5, but the article uses tentative wording such as “may” and “likely.”

The Decoder reported, citing The Information and an internal memo, that OpenAI had reportedly finished pretraining a new model code-named Spud. That is still a reported internal milestone, not the same thing as OpenAI publicly launching a model.

Another article focused on GPT-5.5/Spud explicitly labels its framing as speculation and says there is no official GPT-5.5 release date, model card, or API pricing yet.

Taken together, the third-party sources show that there is a Spud/GPT-5.5 rumor. They do not prove that Spud has been released, that Spud will definitely be called GPT-5.5, or that GPT-5.5 would be a newly pretrained foundation model.

Why the version number does not settle the question

The phrase “new foundation model” usually points to the underlying pretrained model family — not just a product label, routing choice, or incremental model update. That distinction is important here.

A label like GPT-5.5 would not, by itself, prove a new foundation model. OpenAI’s public naming already includes GPT-5.3 and GPT-5.4 within the GPT-5 line, so if GPT-5.5 appears later, the cautious first read would be that it is another GPT-5-series step.

But product names are not technical proof. To make a stronger claim that a model is a new foundation model, outside observers would normally need official evidence such as a model page, API model ID, model card, pricing, release notes, or a formal launch post. At least one current Spud/GPT-5.5 source explicitly says those official release signals have not been announced.

In other words: even if reports of Spud completing pretraining are accurate, that does not automatically establish that OpenAI has publicly released GPT-5.5, or that GPT-5.5 has been confirmed as a new foundation model.

Evidence check

What developers and product teams should do

If you are planning model routing, cost estimates, migration work, or a product roadmap, do not treat GPT-5.5 Spud as a confirmed dependency yet. The safer source of truth is still OpenAI’s official API and product documentation, which in the supplied materials points to GPT-5.4 rather than GPT-5.5.

For low-latency, high-volume workloads, OpenAI’s GPT-5 mini documentation recommends starting with GPT-5.4 mini. For workloads where speed and cost are the main constraints, the GPT-5 nano documentation recommends GPT-5.4 nano.

Spud is worth watching, but it should not be used yet as a production promise, pricing assumption, or capability assumption. The signals that would change the picture are straightforward: an official OpenAI model ID, model card, API pricing page, release note, or product announcement. The current rumor-focused sources do not provide that full official evidence.

Bottom line

The most accurate answer today is: we cannot confirm that GPT-5.5 Spud is a new foundation model — and we cannot even confirm from the supplied official materials that GPT-5.5 Spud has been formally announced by OpenAI.

If OpenAI later adopts the GPT-5.5 name, the existing GPT-5.3/GPT-5.4 naming pattern suggests it should first be understood as a GPT-5-series progression. Whether it is also a new foundation model is a separate technical question that should wait for official documentation, not version-number guesswork or third-party rumor.