OpenAI Retires GPT-4o Entirely — Introduces $100 Pro Plan and GPT-5.3 Instant Mini
GPT-4o is fully retired from all ChatGPT plans as of April 3. OpenAI adds a new $100/month Pro tier for Codex power users and ships GPT-5.3 Instant Mini as the new fallback model.
OpenAI has fully retired GPT-4o from all ChatGPT plans as of April 3, 2026, completing a phased sunset that began on February 13. The company simultaneously shipped GPT-5.3 Instant Mini as a new fallback model and introduced a $100/month Pro plan positioned between Plus and the existing $200 Pro tier.
GPT-4o Is Gone
GPT-4o served as OpenAI's workhorse model from mid-2024 through early 2026 — the first model to unify text, vision, and voice in a single architecture. But usage collapsed after the GPT-5 family launched. According to OpenAI's retirement announcement, only 0.1% of ChatGPT users were still selecting GPT-4o daily by the time the plug was pulled.
The retirement also affects GPT-4.1, GPT-4.1 mini, and o4-mini. Custom GPTs that referenced retired models have been automatically migrated to their closest GPT-5.3 Instant or GPT-5.4 equivalents.
Business, Enterprise, and Edu customers had extended access through April 3 for Custom GPT migration. That window is now closed.
GPT-5.3 Instant Mini: The New Fallback
GPT-5.3 Instant Mini replaces the previous GPT-5 Instant Mini as the model users hit when they exceed their rate limits. OpenAI says it delivers "more natural conversation, stronger writing, and better contextual awareness" compared to its predecessor.
It's a small but meaningful upgrade. Fallback models matter more than most users realize — they're what ChatGPT actually runs during peak hours or after heavy usage. Making the floor higher makes the overall experience more consistent.
The New $100/Month Pro Plan
OpenAI's pricing now has three paid tiers:
| Plan | Price | Key benefit |
|---|---|---|
| Plus | $20/mo | GPT-5.3 Instant, limited GPT-5.4 |
| Pro | $100/mo | Unlimited GPT-5.4, up to 10x Codex usage |
| Pro (original) | $200/mo | GPT-5.4 Pro, highest limits |
The new $100 tier targets developers who need extended Codex sessions but don't need the full $200 Pro package. Unlimited GPT-5.4 access alone makes it a significant step up from Plus.
Yes, having two plans both called "Pro" is confusing. OpenAI hasn't addressed the naming.
What This Means for the Market
OpenAI is consolidating aggressively around the GPT-5 family. Retiring GPT-4o — once the company's most advanced public model — signals that the GPT-5 line has matured enough to handle all workloads without a safety net.
For enterprise users still on Azure OpenAI, Microsoft's retirement timeline follows a similar schedule. Check your deployment configurations if you haven't already.
The competitive implication: Claude and Gemini now compete exclusively against GPT-5.x, not a mixed roster. That raises the bar for everyone.
Our Take
Retiring GPT-4o at 0.1% daily usage is the right call — keeping zombie models running wastes infrastructure and fragments the developer experience. The $100 Pro plan fills a genuine gap; $200/month was too steep for most individual developers, and $20 didn't include enough Codex capacity for serious coding work.
But the naming? Two plans called "Pro" at different price points is a self-inflicted usability wound. OpenAI should fix that before it causes billing confusion at scale.
FAQ
What happens to my Custom GPTs that used GPT-4o? OpenAI has automatically migrated them to GPT-5.3 Instant or GPT-5.4 equivalents. Check your GPT settings to verify which model was assigned and test outputs — behavior may differ slightly.
Is the $100 Pro plan worth it over Plus? If you use Codex regularly, yes. The 10x usage increase and unlimited GPT-5.4 access justify the price difference. If you mainly use ChatGPT for conversation, Plus at $20 remains sufficient.
Does GPT-4o still work via the API? The ChatGPT retirement and API deprecation follow different timelines. Check OpenAI's deprecation schedule for API-specific dates.
How does GPT-5.3 Instant compare to Claude Sonnet 4.6? GPT-5.3 Instant is optimized for speed and conversational quality. Claude Sonnet 4.6 tends to outperform on complex reasoning and coding tasks at a similar price point, though GPT-5.3 has faster response times for general chat.