OpenAI’s next flagship model is no longer a rumor. Sam Altman confirmed in late March 2026 that pretraining for the company’s upcoming frontier model, internally codenamed Spud, finished on March 24. He described it as “a few weeks” from release. As of early April, that window is open.
What OpenAI has actually confirmed
The pretraining is done. Greg Brockman, who returned to OpenAI earlier this year, called it “two years of research” with a “big model feel,” suggesting this is a generational step rather than an incremental update. Altman has been characteristically vague about specifics but clear about timing: the model is coming soon, not in months.
One notable point of confusion is the name. OpenAI has not confirmed whether the release will be called GPT-5.5 or GPT-6. The company’s recent naming history, from GPT-5.2 to GPT-5.4 to various Codex variants, suggests it may opt for a more conservative version number. Regardless of what it is called, the internal codebase marks it as a significant generational leap.
The competitive pressure pushing OpenAI forward
Spring 2026 is shaping up as the most competitive AI model season on record. Anthropic launched Claude Mythos on April 7. Google shipped Gemini 3.1 Ultra with a two-million token context window. As FindSkill.ai noted in its recent analysis of the GPT-6 timeline, waiting risks momentum, which likely accelerated OpenAI’s timeline.
ChatGPT currently holds the largest user base among paid AI subscribers at around 55.2% market share, according to available estimates. Protecting that lead while competitors ship flagship updates is the central challenge OpenAI faces this month.
What GPT-6 will likely need to do
Based on the competitive landscape, matching Gemini’s two-million token context window appears to be table stakes. Surpassing Claude Mythos on general capability benchmarks would be the headline achievement. There is also significant speculation about tighter Codex integration, which could make the main model far more capable for coding workflows than GPT-5.4 currently is.
What this means for users right now
If you are currently using ChatGPT Plus, no action is needed. As noted by analysts covering the release, the upgrade typically rolls out automatically to paid subscribers when it lands. GPT-5.4 remains fully capable for everyday tasks in the meantime.
For anyone evaluating whether to start with ChatGPT or a competitor like Claude or Gemini, this is an unusually awkward moment. Two major are about to receive significant model upgrades within weeks of each other. If timing matters less than current capability, Claude Opus 4.6 leads on coding and analysis today. If you want the largest ecosystem and expect GPT-6 to reset the rankings, sticking with ChatGPT Plus makes sense.
