# OpenAI o3

37 Hacker News commenters recommended OpenAI o3 and 13 criticized it, November 2025 to November 2026. It was recommended in 11 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** People recommend o3 for analytical, medical and coding questions, where it gave fewer wrong or nonsensical answers. The common caveat is that o3 Pro is slow and expensive.

## Why people recommend it

- Strong at identifying places from images (5 comments): "5.5 is significantly worse than o3 at geo-guessing" ([source](https://news.ycombinator.com/item?id=48220221)) "O3 has been excellent at identifying unknown finds" ([source](https://news.ycombinator.com/item?id=48224236)) "also give the correct answer (without web search)" ([source](https://news.ycombinator.com/item?id=48816071))
- Detailed, thorough answers on first try (4 comments): "80% of the thoroughness with 20% of the time, which tended to be better overall" ([source](https://news.ycombinator.com/item?id=47145617)) "o3 is getting it no problem, first try, a simple and reasonable answer" ([source](https://news.ycombinator.com/item?id=45912707)) "after 15 minutes it returned with the answer that had taken me hours" ([source](https://news.ycombinator.com/item?id=48433556))
- Strong at coding and Rust (3 comments): "Since O3, which is the first one that really started to understand Rust's borrow checker in my experience." ([source](https://news.ycombinator.com/item?id=46041677)) "Its the secret sauce behind why the current models are so great at coding" ([source](https://news.ycombinator.com/item?id=48471754)) "o3 was the best at the programming tasks I do" ([source](https://news.ycombinator.com/item?id=47056816))
- Fewer wrong or nonsensical answers (3 comments): "It gives me fewer flat-out wrong or nonsensical responses." ([source](https://news.ycombinator.com/item?id=46617015)) "o3 did a notably better job at a lot of analytical prompts" ([source](https://news.ycombinator.com/item?id=46817468)) "o3 is a much better model for me than GPT5 on any setting" ([source](https://news.ycombinator.com/item?id=45820858))

## What people warn about

- Slow and expensive to run (4 comments): "o3 pro is very slow and expensive" ([source](https://news.ycombinator.com/item?id=46001801)) "o3 actually costs $350 per hour (more than the human price) to achieve tasks at its full 1.5 hour task horizon." ([source](https://news.ycombinator.com/item?id=47812415)) "That's orders of magnitude cheaper than OpenAI o3 and qualitatively in a different league." ([source](https://news.ycombinator.com/item?id=46834136))
- Overhyped claims about its abilities (3 comments): "the meta around AI is not rigorous reporting on the nuance of capabilities but bold claims that are easy to retweet" ([source](https://news.ycombinator.com/item?id=48220166)) "I remember reading breathless threads about how o3 was going to single-handedly end white collar work" ([source](https://news.ycombinator.com/item?id=49527747)) "showed they were not that good at it." ([source](https://news.ycombinator.com/item?id=48297275))
- Overreaches and misreports its methods (2 comments): "then when it responded claimed to recognize the visuals of the mountain" ([source](https://news.ycombinator.com/item?id=48220753)) "o3 was cheating — it pulled location data, then lied about it." ([source](https://news.ycombinator.com/item?id=48220248))

## Alternatives hackers recommend

- [VibeThinker 3B](https://hackerapproved.com/p/vibethinker-3b.md): 28 recommend
- [DeepSeek R1](https://hackerapproved.com/p/deepseek-r1.md): 25 recommend
- [OpenAI o1](https://hackerapproved.com/p/openai-o1.md): 12 recommend

## Where to get it

- [Visit the official site](https://chatgpt.com/)

## Cite as

Hacker Approved, "OpenAI o3", https://hackerapproved.com/p/o3/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).