OpenAI's next major AI model, GPT-6, is likely to arrive between late April and early June 2026, according to credible timelines and statements from company leadership. The company's frontier model, codenamed "Spud," completed pretraining on March 24, 2026, and Sam Altman described it as "a few weeks" away from launch. However, OpenAI has not published an official release date, and an unverified April 14 launch rumor came and went without any announcement from the company .

The silence around GPT-6 contrasts sharply with OpenAI's rapid release cadence for the GPT-5 family. The company shipped five GPT-5 models in under seven months, with GPT-5.4 launching on March 5, 2026. That latest version includes native computer use capabilities and can process a 1-million-token context window, roughly equivalent to 750,000 words in a single conversation . This aggressive iteration strategy is deliberate; OpenAI is gathering real-world data from each GPT-5 generation to refine the agentic behaviors that will define GPT-6.

When Will GPT-6 Actually Launch?

The most credible launch window is April 21 through late May 2026, based on Altman's "few weeks" framing and standard safety evaluation timelines. Frontier AI releases typically undergo 4 to 6 weeks of safety testing, including red teaming, a structured process where researchers probe models for failures before public release. Polymarket traders, who bet on AI milestones, initially priced "GPT-6 by April 30" at 78% probability but trimmed that to roughly 72% after the April 14 rumor failed to materialize. The same traders assign approximately 95% probability to a GPT-6 release by June 30, 2026 .

An unverified leak claimed April 14, 2026 as the launch date and tied it to a "super app" combining ChatGPT, Codex, and the Atlas browser. That date passed without an OpenAI tweet, blog post, or official announcement. This is a useful reminder that AI rumors spread quickly on social media, but true leaks rarely appear as simple text outputs or screenshots. Large language models (LLMs), the AI systems that power ChatGPT, are predictive engines; when asked "When are you released?", they often generate plausible-sounding dates based on patterns in internet text rather than accessing internal company information .

What Features Should You Expect From GPT-6?

OpenAI has not published technical specifications for GPT-6, but the company's public investments and statements point to a clear direction. The most significant leap will likely be in memory and personalization. Sam Altman has described current AI memory as being in the "GPT-2 era" relative to what's coming, a striking claim that suggests persistent memory is where GPT-6 will make its biggest breakthrough. Today's models have a context window that limits how much they remember in a single session, and they struggle to retain information between different conversations unless users explicitly remind them .

GPT-6 is expected to introduce capabilities that fundamentally change how users interact with AI:

• Persistent Memory: The AI might remember details from conversations you had months ago without needing reminders, reducing the burden of re-explaining context.
• Adaptive Learning: It could learn your writing style, coding preferences, or business tone without requiring new instructions every time you interact with it.
• Agentic Workflows: OpenAI is betting heavily on agents, tools that take actions on your behalf rather than simply answering questions, as evidenced by the OpenAI Agents SDK.

This shift from "chatting" to "doing" represents a fundamental change in how AI assistants operate. Instead of starting from scratch with every new conversation, you would be continuing a relationship where you left off. Imagine an AI that knows you prefer Python over JavaScript, or that you always want your marketing emails to have a professional but friendly tone. This reduces the "prompt engineering" burden on users significantly, making AI more intuitive and personalized .

How to Prepare for GPT-6 Without Waiting

If you are building workflows or applications that depend on AI, experts recommend staying flexible rather than waiting for GPT-6 to arrive. Here are practical steps to take now:

• Build on Current Models: Use GPT-5.4 or other available models to develop your workflows today; you can upgrade to GPT-6 quickly once it launches without losing your progress.
• Design for Flexibility: Structure your applications so that swapping in a new model requires minimal code changes, allowing you to adopt GPT-6 as soon as it becomes available.
• Avoid Unverified Claims: Be skeptical of "early access" offers or leaked screenshots claiming to show GPT-6 features; enable multi-factor authentication (MFA) on your OpenAI account and only use official sign-up flows.

The infrastructure supporting GPT-6 is also worth monitoring. OpenAI's Stargate Project signals serious investment in computing power, though the company has not explicitly linked it to a GPT-6 release date. Treat infrastructure announcements as context about scale rather than a countdown timer .

The bottom line is straightforward: GPT-6 is coming, likely within the next 6 to 8 weeks, but OpenAI is not rushing to confirm a date. The company is prioritizing safety testing and real-world data collection from GPT-5 iterations. For developers and businesses, the best strategy is to build with what's available now and stay ready to upgrade when the next frontier model actually arrives.