OpenAI just dropped o3-pro This is their best reasoning model for...
Min Choi@minchoi
79 views
Jun 11, 2025
~1 min read
Advertisement
1
OpenAI just dropped o3-pro
This is their best reasoning model for ChatGPT yet.
Excels at math, science and coding 🤯
10 wild examples & takes:
1. Handles realistic collisions between balls and walls
This is their best reasoning model for ChatGPT yet.
Excels at math, science and coding 🤯
10 wild examples & takes:
1. Handles realistic collisions between balls and walls
0:26
2
2. o3-pro - “it’s smarter. much smarter.”
x.com/morqon/status/…
x.com/morqon/status/…
3
3. o3-pro is wiser and more thoughtful
Derya Unutmaz, MD@DeryaTR_
I have many mind blowing examples of o3-pro outputs, but let me quickly share one. In this instance, I’ve been working with o3-pro to develop immune system 2.0, a modestly ambitious attempt to completely reengineer our immune system 😆 I first asked o3-pro to identify key limitations of our natural immune system and posed the same question to the o3 model.
The responses from o3-pro (the first screenshot) were definitely wiser, more thoughtful, suggesting deeper understanding of the immune system. For example, while the first five issues listed by o3 (second screenshot) highlighted similarly important problems, the last two were issues that could be addressed without completely reengineering the immune system; indeed, its suggested solutions were oriented accordingly. In contrast, o3-pro not only provided more limitations, but each limitation it identified was an extraordinarily critical, intrinsic issue requiring complete reengineering.
In a follow-up post later today, I’ll share the solutions each model proposed. Suffice it to say, while o3 is outstanding, o3-pro was superior in every aspect: innovation, depth, and understanding.
I believe, these sort of qualitative distinctions are not reflected in standard benchmarks, but for me, this real-world comparison is a far better indicator of o3-pro’s capability! I’m incredibly excited about this!
The responses from o3-pro (the first screenshot) were definitely wiser, more thoughtful, suggesting deeper understanding of the immune system. For example, while the first five issues listed by o3 (second screenshot) highlighted similarly important problems, the last two were issues that could be addressed without completely reengineering the immune system; indeed, its suggested solutions were oriented accordingly. In contrast, o3-pro not only provided more limitations, but each limitation it identified was an extraordinarily critical, intrinsic issue requiring complete reengineering.
In a follow-up post later today, I’ll share the solutions each model proposed. Suffice it to say, while o3 is outstanding, o3-pro was superior in every aspect: innovation, depth, and understanding.
I believe, these sort of qualitative distinctions are not reflected in standard benchmarks, but for me, this real-world comparison is a far better indicator of o3-pro’s capability! I’m incredibly excited about this!
4
4. Making word ladder from SPACE to EARTH
Ethan Mollick@emollick
Been playing with o3-pro for a bit. It is quite smart. One problem it solved where every other model has failed is making word ladder from SPACE to EARTH.
(Probably not contamination: the answer is different than the only online answer, which is for EARTH to SPACE in any case)
(Probably not contamination: the answer is different than the only online answer, which is for EARTH to SPACE in any case)
5
5. o3-pro passes the "reindeer" test
x.com/koltregaskes/s…
x.com/koltregaskes/s…
6
6. o3-pro is much better tuned agentic model but uses lot of context
Latent.Space@latentspacepod
🆕It's @openai o3-pro launch day!
our high taste guest tester @benhylak has been previewing for the past week and found an interesting pattern: (link in reply)
o3-pro doesn't noticeably outperform in normal situations, but it's just really, really, REALLY good at consuming ALL your context and using adroitly tools to get EVEN more context. o3-pro is a superlatively much better tuned agentic model than we've ever seen (yes, more than 4 Opus and 2.5 Pro, really — not just "big model smell" but flexing true agentic capability).
our high taste guest tester @benhylak has been previewing for the past week and found an interesting pattern: (link in reply)
o3-pro doesn't noticeably outperform in normal situations, but it's just really, really, REALLY good at consuming ALL your context and using adroitly tools to get EVEN more context. o3-pro is a superlatively much better tuned agentic model than we've ever seen (yes, more than 4 Opus and 2.5 Pro, really — not just "big model smell" but flexing true agentic capability).
7
7. "o3-pro is the slowest and most overthinking model"
8
8. o3-pro can give you game plan to become 50% happier based on your chat history
Yana Welinder@yanabana
o3-pro just dropped.
After 2 mins of reasoning, it gave me a legit game plan to become 50% happier based on my chat history.
What other use cases should I try?
After 2 mins of reasoning, it gave me a legit game plan to become 50% happier based on my chat history.
What other use cases should I try?
9
9. "o3-pro is crazy good"
x.com/Yampeleg/statu…
x.com/Yampeleg/statu…
10
10. PROMPT: A majestic representation of yourself
adi@adonis_singh
grateful to @edwinarbus & @openaidevs for giving me early access to o3-pro, their highest-end reasoning model till date!
I have had a lot of fun solving and creating with o3 pro, here are some cool things I did with it!
(prompt: A majestic representation of yourself)
I have had a lot of fun solving and creating with o3 pro, here are some cool things I did with it!
(prompt: A majestic representation of yourself)
11
If you enjoyed this thread,
Follow me @minchoi and please Bookmark, Like, Comment & Repost the first Post below to share with your friends:
Follow me @minchoi and please Bookmark, Like, Comment & Repost the first Post below to share with your friends:
Min Choi@minchoi
OpenAI just dropped o3-pro
This is their best reasoning model for ChatGPT yet.
Excels at math, science and coding 🤯
10 wild examples & takes:
1. Handles realistic collisions between balls and walls
This is their best reasoning model for ChatGPT yet.
Excels at math, science and coding 🤯
10 wild examples & takes:
1. Handles realistic collisions between balls and walls