# Talkie

19 Hacker News commenters recommended Talkie and 13 criticized it, November 2025 to November 2026. It was recommended in 4 of 13 months.

**Verdict:** People recommend this chatbot-style model, trained only on pre-1931 text, as a distinctive window into the writing and views of its era. Critics warn its training data is contaminated with later text, so results may be unreliable.

## Why people recommend it

- Fun to try and explore (3 comments): "I love this, such a cute little guy." ([source](https://news.ycombinator.com/item?id=47933920)) "best used for cool hacks like this" ([source](https://news.ycombinator.com/item?id=47934100)) "The live-feed is super cool." ([source](https://news.ycombinator.com/item?id=47938837))
- Authentic view of 1930s perspectives (3 comments): "The pre-1931 cutoff date for this model is probably as close to the end of the optimistic age as it's reasonable to get." ([source](https://news.ycombinator.com/item?id=47936735)) "This is so awesome they built this, I've wondered what an LLM only trained on pre-1950s data would look like!" ([source](https://news.ycombinator.com/item?id=47938717)) "It's much more useful for getting vibes / gist of perspectives of the day." ([source](https://news.ycombinator.com/item?id=47950114))
- Useful control baseline for research (2 comments): "It is cool. I find the idea of trying to understand whether these types of models can come up with things like General relativity" ([source](https://news.ycombinator.com/item?id=47928987)) "a great control baseline for studying what's actually learned from pretraining vs. scale" ([source](https://news.ycombinator.com/item?id=47934135))
- Fully open, copyright-free model released (2 comments): "one of the only models fully unencumbered from copyright" ([source](https://news.ycombinator.com/item?id=47933961)) "actually releasing a model counts for a whole lot" ([source](https://news.ycombinator.com/item?id=47930385))

## What people warn about

- Training data contaminated with later text (5 comments): "The fact that the model is contaminated doesn't mean that it won't behave mostly like a model from that era. But it casts doubt on everything done with this model" ([source](https://news.ycombinator.com/item?id=47965026)) "But as they acknowledge later, there are multiple major data leakage issues in their training pipeline" ([source](https://news.ycombinator.com/item?id=47929190)) "This model is contaminated in subtle ways that make me skeptical of the results." ([source](https://news.ycombinator.com/item?id=47931087))
- Hallucinates and drifts into nonsense (5 comments): "it's trying to convince me that he was born in 1839, even after I told it he was born in 1889" ([source](https://news.ycombinator.com/item?id=47948848)) "it hallucinated a film from 1923 that takes place in the Antarctic" ([source](https://news.ycombinator.com/item?id=47945658)) "Then it riffs on that, drifting off into plausible nonsense." ([source](https://news.ycombinator.com/item?id=47930609))
- Repeats biased or offensive views (2 comments): "ask it about its opinion on "coloured people" and you will lose every ounce of skepticism" ([source](https://news.ycombinator.com/item?id=47959296)) "So finally it is settled. The war was about state rights, not about slavery." ([source](https://news.ycombinator.com/item?id=47934074))
- Hard to host; no mobile support (2 comments): "hosting talkie myself was a slog and I had problems using platforms" ([source](https://news.ycombinator.com/item?id=47944992)) "The chat does not work on mobile /tablet" ([source](https://news.ycombinator.com/item?id=47937240))

## 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, "Talkie", https://hackerapproved.com/p/talkie-13b-vintage-language-model/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).