# GPT-2

12 Hacker News commenters recommended GPT-2 and 25 criticized it, November 2025 to November 2026. It was recommended in 6 of 13 months.

**Verdict:** Recommended as an early language model that was impressive for 2019 and foreshadowed later AI. Critics say its output was barely coherent and its withheld release was overhyped.

## Why people recommend it

- Impressive for its time (2 comments): "Just stringing together a few sentences while maintaining logical coherence was very impressive at the time." ([source](https://news.ycombinator.com/item?id=48423452)) "It was still impressive for 2019, but yeah, it was nothing to worry about." ([source](https://news.ycombinator.com/item?id=47684702))
- Foreshadowed later model behavior (2 comments): "that generation of models was around the point where we were getting pleasantly surprised by the behaviors of models" ([source](https://news.ycombinator.com/item?id=45948372)) "Obviously GPT-2 wasn’t great at it, but the writing was on the wall quite literally." ([source](https://news.ycombinator.com/item?id=48422439))
- Interesting despite being useless (2 comments): "smaller versions of the model that were interesting but pretty useless" ([source](https://news.ycombinator.com/item?id=48471411)) "Practically useless, but interesting." ([source](https://news.ycombinator.com/item?id=50047827))

## What people warn about

- Output barely coherent, Markov-chain-like (6 comments): "What I'm getting from it looks like it could just be arbitrary sentence and paragraph fragments from the Internet, pasted together Markov-chain like." ([source](https://news.ycombinator.com/item?id=49885465)) "GPT-2 wasn't capable of anything other than imitating a stroke victim." ([source](https://news.ycombinator.com/item?id=49623288)) "GPT-2 could do those things about as convincingly as a Markov chain." ([source](https://news.ycombinator.com/item?id=49634697))
- Withheld release seen as overhyped (4 comments): "OpenAI's qualms weren't with making the dangerous models usable, they were with making the model usable without paying them." ([source](https://news.ycombinator.com/item?id=47689259)) "when it was the first time OpenAI restricted access to the models citing "humanity is not ready for it"" ([source](https://news.ycombinator.com/item?id=48242225)) "We cared about misuse right up until it became apparent that was profit to be had" ([source](https://news.ycombinator.com/item?id=47795966))
- Far worse than newer models (2 comments): "vanilla transformers like GPT 2 are ridiculously inefficient in comparison" ([source](https://news.ycombinator.com/item?id=50000121)) "it sucks much much worse than GPT-2 did with dungeon text adventures" ([source](https://news.ycombinator.com/item?id=46575984))
- Called useless and unimpressive (3 comments): "GPT-2 was effectively useless because it was only trained from links posted on Reddit" ([source](https://news.ycombinator.com/item?id=48467400)) "smaller versions of the model that were interesting but pretty useless" ([source](https://news.ycombinator.com/item?id=48471411)) "It was bullshit in 2019 it's bullshit now too." ([source](https://news.ycombinator.com/item?id=48468809))

## Alternatives hackers recommend

- [Claude Opus](https://hackerapproved.com/p/claude-opus.md): 3331 recommend
- [OpenAI GPT](https://hackerapproved.com/p/gpt-5-5.md): 1205 recommend
- [Claude Sonnet](https://hackerapproved.com/p/claude-sonnet.md): 854 recommend
- [DeepSeek V4 Flash](https://hackerapproved.com/p/deepseek-v4-flash.md): 828 recommend
- [Fable](https://hackerapproved.com/p/fable.md): 823 recommend

## Cite as

Hacker Approved, "GPT-2", https://hackerapproved.com/p/gpt-2/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).