ChatGPT has more than 900 million weekly active users and over 50 million subscribers as of April 2026, yet OpenAI just released its newest model, GPT-5.5, a mere six weeks after the company debuted GPT-5.4. That's not a typo. The company that reshaped how millions of people work is now shipping frontier AI models faster than most software companies push bug fixes. If you're trying to keep up with what ChatGPT can actually do today, which tier you need, or whether OpenAI is still leading the pack, this is everything that's changed in April 2026.
| What Changed | Who It Affects | Why It Matters |
|---|---|---|
| GPT-5.5 launched April 23 | Plus, Pro, Business, Enterprise users | Faster agentic coding, better tool use, same latency as 5.4 but fewer tokens per task |
| API available April 24 | Developers | $5/$30 per 1M tokens (input/output), 2x GPT-5.4 pricing but more efficient |
| New $8/month Go tier expansion | Budget users in cost-sensitive markets | 122M paid users projected by end of 2026, up from 47M in 2025 |
| Ads now in Free + Go tiers | Non-paying and low-tier users | OpenAI targeting $2.4-2.5B ad revenue in 2026, up to $100B by 2030 |
| Pro tier split: $100 and $200 | Power users | Pro $100 launched April 9 to compete with Claude Max at same price |
OpenAI released GPT-5.5 on April 23, 2026, calling it "our smartest and most intuitive to use model yet". The headline features include stronger multi-step reasoning, better tool use, agentic coding improvements, and the ability to "give GPT-5.5 a messy, multi-part task and trust it to plan, use tools, check its work, navigate through ambiguity, and keep going". OpenAI president Greg Brockman called it "a new class of intelligence" during the press briefing.
The model is available to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, with GPT-5.5 Pro available to Pro, Business, and Enterprise tiers. For API developers, pricing is set at $5 per 1M input tokens and $30 per 1M output tokens, with GPT-5.5 Pro priced at $30/$180. That's roughly double the cost of GPT-5.4, but OpenAI says "it is both more intelligent and much more token efficient".
Benchmark claims show improvements on Terminal-Bench 2.0 and coding evals, though independent tests from Tom's Guide found GPT-5.5 lost in all 7 categories against Claude Opus 4.7, with the site praising speed but criticizing hallucination tendencies. OpenAI countered with usage stats: 4 million active Codex users and 9 million paying business users. The subtext is clear. They're racing Anthropic, and the gap is narrower than either company wants to admit.
OpenAI now operates seven pricing tiers, and the structure changed twice in April alone. The big shift is volume over premium. OpenAI expects its paid ChatGPT user base to grow from around 47 million in 2025 to about 122 million by 2026, driven by a new $8 per month plan that could bring in over 100 million users. Meanwhile, the $20 ChatGPT Plus tier may witness a sharp decline as users downgrade or new subscribers choose affordability.
Here's what the tiers look like as of late April 2026:
The Go tier is the tell. Average revenue per user is expected to decline sharply, from about $23 per month to under $12. To compensate, OpenAI's ad pilot generated over $100 million in annualized revenue within just weeks of launch, with the company projecting around $2.4-$2.5 billion in ad revenue in 2026.
| Provider | Model | Price | Strengths | Weaknesses |
|---|---|---|---|---|
| OpenAI | GPT-5.5 | $20/mo (Plus) | Broadest ecosystem, Codex, Advanced Voice, Sora video, 60+ integrations | Lost head-to-head tests vs Claude 4.7, API costs doubled |
| Anthropic | Claude Opus 4.7 | $20/mo (Pro), $100/mo (Max 5x) | 1M-token context on Max vs ChatGPT Pro's 256K, 64.3% on SWE-Bench Pro vs GPT-5.5's 58.6% | Banned third-party agents April 4, smaller ecosystem |
| Gemini 3.1 Pro | $20/mo | Free tier most generous, seamless Google Workspace integration | Inconsistent quality, lagging on reasoning benchmarks | |
| xAI | Grok (SuperGrok) | $30/mo | Bundled with X Premium, real-time X data access | Distant fourth, limited use outside X ecosystem |
The $20/month convergence isn't coincidence. OpenAI established this price point in February 2023, and competitors matched it to avoid appearing more expensive, with actual computational cost estimated at $5-10/month for typical usage. Every provider is making 50-75% gross margins on subscriptions. The real competition is happening in feature depth, not pricing.
Anthropic banned third-party agents from Claude Pro and Max on April 4; five days later, OpenAI launched Pro $100 priced identically to Claude Max 5x. If you were running agents on Claude Code and got cut off, OpenAI built you a landing pad. That's not product development. That's a ground war.
OpenAI shipped GPT-5.2 in December 2025, GPT-5.4 in March, and GPT-5.5 in late April, a sub-two-month rhythm. Fortune noted that AI model launches are increasingly evolving through continuous, incremental updates, and chief scientist Jakub Pachocki said on the April 23 call that users should expect "pretty significant improvements in the short term, extremely significant improvements in the medium term".
This is a structural shift. Model releases used to be events. GPT-3 to GPT-4 took 15 months. Now we're seeing point releases in weeks. That's not just faster iteration. It's a different product philosophy, and it's forcing every downstream user (enterprises, developers, power users) to rethink integration strategies. You can't treat every release as a one-off when they're shipping every six weeks.
Meanwhile, OpenAI is expanding into verticals. ChatGPT for Clinicians launched as a free U.S. version for verified clinicians, with clinical search, citations, and CME credits. GPT-Rosalind for life sciences research launched April 16. Codex got browser use for local development servers, letting Codex click through rendered UI and reproduce visual bugs inside the app. And OpenAI announced an expanded strategic partnership with AWS, bringing GPT-5.5 and other frontier models to Amazon Bedrock.
The subtext again: OpenAI is no longer just a model company. It's building vertical products, enterprise infrastructure, and advertising infrastructure simultaneously. That's the playbook of a company preparing to go public, not a research lab.
Two moves matter more than the model itself. First, OpenAI launched GPT-5.5 and other frontier models on Amazon Bedrock and brought Codex to AWS, ending the Microsoft exclusivity that defined OpenAI's early enterprise strategy. Customers can now run OpenAI models inside their existing AWS environments with the security and compliance controls they already have. That's a direct shot at Google's Vertex AI and Anthropic's Bedrock partnership.
Second, the advertising ramp. Ads launched on Free tier in February 2026 and expanded to Go tier. Ads are shown using information that stays within ChatGPT and is not shared with advertisers, including what you're discussing and past chats if personalization is enabled, though advertisers only receive aggregated performance data. OpenAI projects ad revenue could grow to almost $100 billion annually by 2030. For context, Google's search ad business did $175 billion in 2025. OpenAI thinks it can build half of Google's search empire in four years.
The Bank of New York has been testing GPT-5.5, and CIO Leigh-Ann Russell said in the April 23 briefing that "the response quality and impressive hallucination resistance" matter for a highly regulated institution, calling it "a step change with this model". That's enterprise validation, and it's the kind of signal OpenAI needs as it races toward an IPO reportedly targeted for late 2026 or 2027.
Yes. The Free tier gives you GPT-5.3 Instant with 10 messages per 5 hours, then falls back to a lighter model. As of February 2026, Free users in the US see ads at the bottom of responses.
GPT-5.5 offers better multi-step reasoning, stronger tool use, and agentic coding improvements at the same per-token latency as GPT-5.4. However, API pricing doubled ($5/$30 per 1M tokens vs $2.50/$15), though OpenAI claims it uses fewer tokens per task.
Plus at $20/month removes ads and gives full feature access including GPT-5.5, Deep Research, Sora, and Codex. Go at $8/month still shows ads and has limited features. If you use ChatGPT for work, the $12 difference is easy to justify.
Claude Opus 4.7 won head-to-head tests in 7 out of 7 categories per Tom's Guide and leads on some coding benchmarks (64.3% vs 58.6% on SWE-Bench Pro). ChatGPT has a broader ecosystem with Advanced Voice, Sora, and 60+ integrations. Both are priced at $20/month for standard tiers.
OpenAI product head Nick Turley said pricing will "significantly evolve" and compared unlimited plans to "unlimited electricity." Analysts predict Plus could rise to $25-30/month by late 2026 or early 2027, though no official increase has been announced.
Launched April 9, 2026, Pro $100 offers the same models as Pro $200 (including GPT-5.5 Pro) but at 5x Plus usage limits instead of 20x. It's targeted at users hitting Plus caps but not needing the full $200 tier, and is priced to compete directly with Claude Max at $100/month.