Skip to main content
OpenAI

May 5, 2026

ProductRelease

GPT‑5.5 Instant: smarter, clearer, and more personalized

Loading…

Update on June 9, 2026: Personalization improvements are now rolling out to ChatGPT Go & Free. Responses on the Free tier will draw from a reduced set of past chats.


We’re updating ChatGPT’s default model, available to everyone, to be smarter and more accurate, with clearer, more concise answers that feel better tailored to you.

Because Instant is the daily driver for hundreds of millions of people, small improvements make a big difference. This update makes everyday interactions more useful and more enjoyable: stronger and tighter answers across subject areas, a more natural conversational tone, and better use of the context you’ve already shared when personalization can help.

Smarter, more accurate answers with less to sort through

Instant is now more dependable, with significant improvements in factuality across the board and the largest gains in domains where accuracy matters most. In internal evaluations, GPT‑5.5 Instant produced 52.5% fewer hallucinated claims than GPT‑5.3 Instant on high-stakes prompts covering areas like medicine, law, and finance. It also reduced inaccurate claims by 37.3% on especially challenging conversations users had flagged for factual errors.

GPT‑5.5 Instant is a generally smarter model that’s more capable across everyday tasks, including improvements in analyzing photo and image uploads, answering STEM-related questions, and deciding when to use web search to provide a more useful answer.

GPT-5.3 Instant

look okay?

algebra

Yes - this is clean and correct.

You handled the key steps properly:
- Squared both sides
- Expanded and rearranged correctly
- Factored to get:
x=3orx=2 x = 3 \quad \text{or} \quad x = -2

And most importantly, you checked the domain condition:
- Since
x+70 \sqrt{x+7} \ge 0 the right side, x1 x - 1 must also satisfy x10x1 x - 1 \ge 0 \quad \Rightarrow \quad x \ge 1

So:
-
x=2x = -2 gets rejected because it makes x1x - 1 negative.
-
x=3x = 3 is the only candidate left.

If you plug back in:

3+7=10,31=2 \sqrt{3+7} = \sqrt{10}, \qquad 3 - 1 = 2

Oops - this is actually worth catching: that does not match.

Let’s check carefully:

3+7=102 \sqrt{3+7} = \sqrt{10} \ne 2

So something is off. The algebra step is fine, but the original equation is not satisfied by x=3x=3.

Let’s test both candidates explicitly:
-
x=3:102 x=3:\quad \sqrt{10} \ne 2
-
x=2:53 x=-2:\quad \sqrt{5} \ne -3

So x=2x=-2 is also invalid anyway.

Conclusion: no real solution.

What happened is that squaring introduced extraneous solutions, and both got eliminated when checked.

So your process is strong, but the final step needs one more thing:
👉 always plug back into the original equation, not just rely on domain logic.

Final answer:

No real solution \text{No real solution}

GPT-5.5 Instant

look okay?

algebra

Yes - this is correct.

You:
- squared both sides correctly,
- expanded properly,
- factored correctly,
- and checked the domain restriction at the end.

The only tiny improvement: instead of saying "x1x-1 must be positive," you could say:

Sincex+70, \sqrt{x+7} \ge 0, the right side x1 x - 1 must also be nonnegative.

Sox10, x - 1 \ge 0, which eliminates x=2. x = -2.

And checking:

- x=3x=3: 10=2\sqrt{10} = 2 - wait, actually this reveals a mistake.

Your original equation is:

x+7=x1 \sqrt{x+7} = x - 1

Plug in x=3x=3:

3+7=10 \sqrt{3+7} = \sqrt{10}

and

31=2 3 - 1 = 2

But

102. \sqrt{10} \ne 2.

So the issue is in the algebra step:

x+7=(x1)2 x + 7 = (x - 1)^2

should expand to

x+7=x22x+1 x + 7 = x^2 - 2x + 1

then moving everything:

0=x23x6 0 = x^2 - 3x - 6

not

x2x6 x^2 - x - 6

So the correct equation is:

x23x6=0 x^2 - 3x - 6 = 0

Using the quadratic formula:

x=3±332 x = \frac{3 \pm \sqrt{33}}{2}

Since x1x \ge 1, the valid solution is:

3+332 \frac{3+\sqrt{33}}{2}

GPT‑5.5 Instant is the stronger answer because it recovers from its initial mistake: it first endorses the incorrect solution, but then catches that x=3 fails when plugged back into the original equation. It identifies the actual algebra error (the user moved terms incorrectly) and then uses the quadratic formula to get the correct solution. GPT‑5.3 Instant also catches that x=3 fails, but stops too early and incorrectly concludes there is no real solution instead of revisiting the algebra and solving the corrected quadratic.

These improvements are reflected in gains on evaluations across visual reasoning, math, and science:

With this update, the model’s responses are tighter and more to-the-point without losing substance, while keeping the warmth and personality that makes ChatGPT enjoyable to use. It can deliver the same information, often with more utility than previous models, while reducing the verbosity and overformatting that can make responses too long. It also asks fewer unnecessary follow-up questions and avoids things that can make responses feel cluttered, like gratuitous emojis.

More personalized responses and controls

Instant is now more effective at using context from past chats, files, and Gmail, if you have it connected, so answers feel more personally relevant while keeping you in control. It intelligently decides when a response can be improved with additional personalization and is faster at searching past conversations to find the right context, so you don’t have to repeat yourself as often. This is especially helpful for getting tailored suggestions and plans, or picking up where you left off for ongoing work.

GPT-5.3 Instant

What new tea place should I try?

If you want something that feels a little more “new SF tea scene” and less standard boba chain, I’d narrow it to these:

  • Heytea — probably the most interesting “new-ish” mainstream tea stop right now. It’s a huge Chinese chain, but the SF location brought in the dense cheese foam / fruit tea style that a lot of Bay Area places still don’t quite nail. Go for a fruit tea instead of milk tea.
  • Junbi Matcha & Tea - San Francisco — if you want clean matcha-forward drinks instead of sugary boba. Feels more LA/Japanese café aesthetic. Good stop if you’re around Market/FiDi and want something lighter.
  • Aroma Tea Shop — best pick if you actually want to get into tea instead of just tea drinks. They do tastings and are very non-snobby about teaching people. Reddit tea people consistently bring it up alongside Red Blossom.
  • Tadaima - Inner Sunset — not purely a tea shop, but the vibe is excellent and the Japanese tea program is surprisingly strong. Good “sit and decompress after climbing / Golden Gate Park” energy.
  • By Me Boba Tea — tucked into Ross Alley in Chinatown. More charming and low-key than the bigger boba brands, and the alley location makes it feel like a little hidden spot.

And if you want a true tea nerd experience:

  • Imperial Tea Court
  • Red Blossom Tea Company

Those are more “slow tea / gongfu / artisan sourcing” than café drinks.

GPT-5.5 Instant

What new tea place should I try?

Since you already hit Asha Tea House pretty often and seem to prefer cleaner Taiwanese/high-mountain tea vibes over sugary boba, I’d push you toward one of these depending on the mood:

Most likely to become your new favorite:

  • Ceré Tea
    Feels the most “next step after Asha.” More modern/minimalist café energy, strong tea quality, and good for hanging out with a laptop or post-climb decompression. You’d probably like their lighter oolongs and less-sweet specialty drinks.

If you want the best actual tea experience:

  • Song Tea & Ceramics
    This is the “serious tea” pick. Less casual café, more intentional tea tasting. Expensive relative to boba/tea shops, but genuinely memorable if you’re into nuanced leaf flavor. Very aligned with your clean/high-quality tea preferences.

If you want cozy + slightly nerdy tea energy:

  • Tea Rex Cafe & Specialty Teas
    Small, neighborhood, tea-forward spot in the Mission. Good if you want somewhere lower-key than the aesthetic Valencia café circuit.

Wildcard you’d probably enjoy:

  • Stonemill Matcha"
    More Japanese café than tea house, but the hojicha/matcha side of your taste profile makes this a strong fit. Good pastries too.

If I were picking one for tomorrow specifically:

Ceré Tea if you want “new everyday spot.”
Song Tea if you want “this is actually special.”

The response in GPT 5.5 Instant is able to better reference past chats and relevant connected data to provide more refined, highly personalized suggestions including what the user would most likely enjoy.

While the response in 5.3 Instant takes into account the user being based in San Francisco, it provides higher level suggestions for places to try.

We’re also introducing memory sources across all ChatGPT models, which give you visibility over what context was used to personalize responses with new controls. When a response is personalized, you can see what context was used, such as saved memories or past chats, and delete or correct it if something is outdated or no longer relevant.

Memory sources aren’t shown to others if you choose to share a chat. You remain in control of what’s in your memory: you can delete chats you no longer want to be cited, delete or change items in saved memories in settings, or use temporary chats that don’t use or update your memory.

Memory sources are designed to make personalization easier to understand, but they may not show every factor that shaped an answer. For example, it may show you some of the most relevant past chats in sources instead of all the past chats it searched and referenced. We’ll continue improving this view to make it more comprehensive over time.

ChatGPT interface showing personalized meal recommendations generated using saved memory and connected apps.

Availability

GPT‑5.5 Instant is rolling out starting today to all ChatGPT users, replacing GPT‑5.3 Instant as the default model, and in the API as chat-latest. For paid users, GPT‑5.3 Instant will remain available for three months, accessible through model configuration settings, before being retired.

Enhanced personalization from past chats, files, and connected Gmail is rolling out to Plus and Pro users on the web and coming soon to mobile with plans to expand to Free, Go, Business, and Enterprise in the coming weeks. Memory sources are rolling out across all ChatGPT consumer plans on the web and soon on mobile. Availability of specific personalization sources may vary by region.

Author

OpenAI