Is GPT-6 Sol coming? OpenAI may usher in a "release frenzy" this week.
OpenAI may be brewing a massive product update.
The first to send out signals was Tibo Sottiaux, Head of Product at OpenAI. He posted on X that, "The number of products and features released this week will reach the level you might have originally expected to see at DevDay 2025."
Shortly afterwards, OpenAI CEO Sam Altman also made a rare direct advance preview: "There will be a major version release this week, and more blockbuster releases will follow during DevDay."
As soon as this news came out, people were speculating: Is GPT-6 Sol coming?
An unofficial account that regularly updates Codex stated directly: "GPT-6 Sol will most likely be released on Thursday."
But this news has not been officially confirmed by OpenAI so far.
However, it seems that OpenAI has already started the gray-scale test for GPT-6 Sol.
Netizen Chetaslua@chetaslua said that when some users were using GPT-5.6 Sol, their background was already routed to GPT-6 Sol. He also stated bluntly that GPT-5.6 Sol focuses heavily on RL and efficiency.
"Sol 6 has invested very deeply in Reinforcement Learning (RL) and the effect is excellent. Dude, it's really extremely fast. OpenAI is leading in efficiency, and the gap compared to Anthropic is further widening. I will soon share the comparison between Sol 6 and Opus 5.2/5.1."
Another netizen also said that there are quite a lot of users included in the gray-scale test. "A large number of users have reported that they have been routed to GPT-6 Sol when using GPT-5.6 Sol. The initial experience feedback is generally positive, and it is said to be extremely fast! It's time for me to test it myself."
Another noteworthy signal is the withdrawal of GPT-5.5.
The official ChatGPT account announced that starting from October 14, GPT-5.5 will no longer be available in ChatGPT, ChatGPT Work and Codex, and users need to migrate to updated models.
This means that OpenAI is pushing users to migrate to the new model system.
This is also regarded by many netizens as OpenAI making room for the next-generation model.
Although this information has not been officially confirmed by OpenAI, after multiple signals are superimposed, a trend is becoming obvious: OpenAI may be preparing for the next-stage model and product ecosystem transition.
GPT-5.5 is gradually phased out;
GPT-5.6 Sol has become the current main model;
GPT-6 Sol is suspected to be in testing;
A large-scale product release will arrive this week;
OpenAI seems to be completing a generation replacement at the model and product level.
It is also worth noting that the protagonist of this "release week" may not be limited to models.
If you shift your focus away from GPT-6 Sol, OpenAI's product feature releases have actually been very intensive in the past week.
On September 8, OpenAI launched ChatGPT Images 2.5, which not only updated the image model, but also added a series of product features such as Sketch, templates, and image comment editing; on the API side, two models GPT-Image-2.5 Flare and Sunburst were launched simultaneously.
And on September 10, OpenAI launched Agents API, GPT-Live-1, Data agent, and ChatGPT for Financial Services for the financial industry consecutively within one day. These products cover intelligent agent infrastructure, real-time voice, enterprise data analysis and vertical industry applications respectively, almost spanning several of the most important product lines of current AI products.
From this perspective, the "products and features" mentioned by Tibo may be even more interesting.
OpenAI is advancing simultaneously in models, multimodality, intelligent agent infrastructure, enterprise data and industry applications. Even if GPT-6 Sol takes up most of the attention, it is very likely just part of this round of updates.
What's more, some netizens believe that after DevDay, OpenAI may enter the Agent era. ChatGPT serves as the user entry, Codex explores software development agents, and tool calling and automation capabilities further expand the model boundary...
In other words, no matter whether the final release is a new model or a series of product capability upgrades, OpenAI is sending a clear signal: The focus of AI competition in the next stage may be shifting from "making the model smarter" to "whether the model can truly complete complex tasks".
And the answer may come this week!
So what about you, what do you think of the upcoming OpenAI "crazy release week"? Feel free to leave a message and share your thoughts!
References:
https://x.com/thsottiaux/status/2099744972195131850
https://x.com/sama/status/2099872600977760451
https://x.com/Codexresets_/status/2099893141247815869
https://x.com/ChatGPT/status/2099954190600876533
https://x.com/chetaslua/status/2099861782710071424
This article is from the WeChat Official Account "Synced" (ID: almosthuman2014), written by AI-focused contributors, and published with authorization from 36Kr.