{"slug": "talkie", "name": "Talkie", "category": "ai-models", "category_name": "AI models", "description": "", "recommenders": 3, "commenters": 9, "warners": 0, "score": 4.408, "rising": 0.35, "first": "2026-04-27", "latest": "2026-09-03", "why": "Talkie[1] presents an interesting case for it being at least possible", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 2, "neutral": 7, "compare": 0, "warn": 0}, "spark": [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "Talkie[1] presents an interesting case for it being at least possible", "kind": "", "icon": true, "domain": "talkie-lm.com", "github_repo": null, "official_url": "https://talkie-lm.com", "pricing": "", "aliases": ["Talkie", "talkie"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 792866256, "item_id": 49554141, "url": "https://news.ycombinator.com/item?id=49554141", "by": "chaosharmonic", "date": "2026-09-03", "version": null, "stance": "recommend", "quote": "Talkie[1] presents an interesting case for it being at least possible", "excerpt": "But how much of that synthetic data still ultimately derives from non-open sources? You'd still have to ask what a clean room implementation ultimately is, depending on how granular or aggressive a large publisher wanted to get about it. That said, I don't necessarily disagree with you. Talkie[1] presents an interesting case for it being at least possible to do this entirely on public domain material. But even that u", "context": "", "use_case": "", "switched_from": "", "story": "K2 Horizon: A connected fleet of six open models"}, {"id": 767058912, "item_id": 47941182, "url": "https://news.ycombinator.com/item?id=47941182", "by": "Nition", "date": "2026-04-28", "version": null, "stance": "recommend", "quote": "You can try it here: https://talkie-lm.com/chat", "excerpt": "You can try it here: https://talkie-lm.com/chat There's a few minutes' wait time, but it's working.", "context": "", "use_case": "Chatting with 1930 vintage model", "switched_from": "", "story": "Talkie: a 13B vintage language model from 1930"}, {"id": 766900048, "item_id": 47931253, "url": "https://news.ycombinator.com/item?id=47931253", "by": "stbullard", "date": "2026-04-28", "version": null, "stance": "recommend", "quote": "I love this.", "excerpt": "USER: What will the future hold for computers? TALKIE-1930: Computers in the future may be employed in offices where calculations are required to be made, and where the nature of the business does not demand a very high degree of knowledge. The occupation will be a steady one, and the remuneration will be sufficient to afford a comfortable livelihood. The hours of attendance will probably be from ten to six. USER: an", "context": "", "use_case": "vintage language model", "switched_from": "", "story": "Talkie: a 13B vintage language model from 1930"}, {"id": 752706144, "item_id": 47044134, "url": "https://news.ycombinator.com/item?id=47044134", "by": "arach", "date": "2026-02-16", "version": null, "stance": "recommend", "quote": "Parakeet and incredibly fast and made for devs", "excerpt": "https://usetalkie.com - Parakeet and incredibly fast and made for devs", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Free Alternative to Wispr Flow, Superwhisper, and Monologue"}]}], "caveats": [], "who": [], "tags": [], "use_cases": ["vintage-data llm", "pre-1930 text llm experiment", "chatting with 1930 vintage model", "vintage language model"], "compared": [], "makers": [{"id": 752706449, "item_id": 47044153, "url": "https://news.ycombinator.com/item?id=47044153", "by": "arach", "date": "2026-02-16", "version": null, "stance": "recommend", "quote": "built my own but with a bit more emphasis on speed", "excerpt": "Handy's great! I find the latency to be just a bit too much for my taste. Like half the people on this thread, built my own but with a bit more emphasis on speed https://usetalkie.com", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Free Alternative to Wispr Flow, Superwhisper, and Monologue"}], "threads": [{"url": "https://news.ycombinator.com/item?id=47927903", "title": "Talkie: a 13B vintage language model from 1930", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47040375", "title": "Show HN: Free Alternative to Wispr Flow, Superwhisper, and Monologue", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=49551760", "title": "K2 Horizon: A connected fleet of six open models", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49317760", "title": "What happens when an LLM never sees material beyond fifth grade?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49181083", "title": "Position: LLMs Can't Jump", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48957779", "title": "GPT-5.6 used a prompt to close a 30-year gap in convex optimization", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {}, "2026-04": {"neutral": 2, "recommend": 1}, "2026-05": {"neutral": 1}, "2026-06": {}, "2026-07": {"neutral": 2}, "2026-08": {"neutral": 2}, "2026-09": {"recommend": 1}, "2026-10": {}}, "versions": [], "version_unnamed": 10, "family_ref": null, "runs_on": null, "lines": [], "quotes": [{"id": 792866256, "item_id": 49554141, "url": "https://news.ycombinator.com/item?id=49554141", "by": "chaosharmonic", "date": "2026-09-03", "version": null, "stance": "recommend", "quote": "Talkie[1] presents an interesting case for it being at least possible", "excerpt": "But how much of that synthetic data still ultimately derives from non-open sources? You'd still have to ask what a clean room implementation ultimately is, depending on how granular or aggressive a large publisher wanted to get about it. That said, I don't necessarily disagree with you. Talkie[1] presents an interesting case for it being at least possible to do this entirely on public domain material. But even that u", "context": "", "use_case": "", "switched_from": "", "story": "K2 Horizon: A connected fleet of six open models"}, {"id": 789085648, "item_id": 49317853, "url": "https://news.ycombinator.com/item?id=49317853", "by": "krackers", "date": "2026-08-16", "version": null, "stance": "neutral", "quote": "A similar project (LLM trained only on vintage material)", "excerpt": "A similar project (LLM trained only on vintage material): https://talkie-lm.com/introducing-talkie", "context": "", "use_case": "Vintage-data LLM", "switched_from": "", "story": "What happens when an LLM never sees material beyond fifth grade?"}, {"id": 787082224, "item_id": 49192639, "url": "https://news.ycombinator.com/item?id=49192639", "by": "handoflixue", "date": "2026-08-05", "version": null, "stance": "neutral", "quote": "They did that with \"Talkie\", a model trained on 1930 and before.", "excerpt": "They did that with \"Talkie\", a model trained on 1930 and before. It had the ability to assemble crude Python programs, but there were definitely some leaks in the training data (it knew about stuff like World War 2) so not 100% perfect. Still seems like a pretty reasonable \"proof of concept\" https://didof.dev/blog/talkie-1930-llm-reasoning/ seems like a decent overview", "context": "", "use_case": "Pre-1930 text LLM experiment", "switched_from": "", "story": "Position: LLMs Can't Jump"}, {"id": 783345616, "item_id": 48959101, "url": "https://news.ycombinator.com/item?id=48959101", "by": "ben_w", "date": "2026-07-18", "version": null, "stance": "neutral", "quote": "Even really rubbish models, e.g. talkie", "excerpt": "The infinite monkey theorem assumes random distribution of symbols*. Given the tokenizers have a vocabulary in the 10k-100k range, \"a million attempts\" will generally still only get the first token of the answer correct. Even really rubbish models, e.g. talkie, the \"what if we only use pre-1930s data to train a model?\"** model, had to be almost all the way to the right answer to reach the really low HumanEval pass@10", "context": "", "use_case": "", "switched_from": "", "story": "GPT-5.6 used a prompt to close a 30-year gap in convex optimization"}, {"id": 782902736, "item_id": 48931421, "url": "https://news.ycombinator.com/item?id=48931421", "by": "red75prime", "date": "2026-07-16", "version": null, "stance": "neutral", "quote": "A model pretrained on texts with 1930 data cutoff can solve a few programming problems", "excerpt": "> ML is as thick as plankton by this definition ...during autoregressive pretraining ([2]). A model pretrained on texts with 1930 data cutoff can solve a few programming problems when given a few examples. Its success rate is understandably much worse. [1] https://talkie-lm.com/introducing-talkie [2] That is when a model starts from a blank state that can be described in a few kilobytes.", "context": "", "use_case": "", "switched_from": "", "story": "The Economics of Recursive Self-Improvement [pdf]"}, {"id": 769430480, "item_id": 48089405, "url": "https://news.ycombinator.com/item?id=48089405", "by": "mrtesthah", "date": "2026-05-10", "version": null, "stance": "neutral", "quote": "This LLM trained only and entirely on pre-1930s texts was able to code Python programs when given only a short example", "excerpt": ">Coding, one of the most popular uses cases today, would not be great if it say only understood java to a version from years ago etc. This LLM trained only and entirely on pre-1930s texts was able to code Python programs when given only a short example: https://talkie-lm.com/introducing-talkie", "context": "", "use_case": "", "switched_from": "", "story": "Local AI needs to be the norm"}, {"id": 767058912, "item_id": 47941182, "url": "https://news.ycombinator.com/item?id=47941182", "by": "Nition", "date": "2026-04-28", "version": null, "stance": "recommend", "quote": "You can try it here: https://talkie-lm.com/chat", "excerpt": "You can try it here: https://talkie-lm.com/chat There's a few minutes' wait time, but it's working.", "context": "", "use_case": "Chatting with 1930 vintage model", "switched_from": "", "story": "Talkie: a 13B vintage language model from 1930"}, {"id": 767052192, "item_id": 47940762, "url": "https://news.ycombinator.com/item?id=47940762", "by": "Nition", "date": "2026-04-28", "version": null, "stance": "neutral", "quote": "I'm trying to convince TALKIE about computers.", "excerpt": "Continuing this conversation, I'm trying to convince TALKIE about computers. USER: I wonder if we could build a calculating machine that would be close enough to perfectly reliable that we would not have to verify the results. For example, you may have seen Leibniz's work showing that all numbers can be represented with just 0 and 1, or George Boole's work on algebra with just two values. Imagine we create an electri", "context": "", "use_case": "", "switched_from": "", "story": "Talkie: a 13B vintage language model from 1930"}, {"id": 766900048, "item_id": 47931253, "url": "https://news.ycombinator.com/item?id=47931253", "by": "stbullard", "date": "2026-04-28", "version": null, "stance": "recommend", "quote": "I love this.", "excerpt": "USER: What will the future hold for computers? TALKIE-1930: Computers in the future may be employed in offices where calculations are required to be made, and where the nature of the business does not demand a very high degree of knowledge. The occupation will be a steady one, and the remuneration will be sufficient to afford a comfortable livelihood. The hours of attendance will probably be from ten to six. USER: an", "context": "", "use_case": "vintage language model", "switched_from": "", "story": "Talkie: a 13B vintage language model from 1930"}, {"id": 766866081, "item_id": 47929130, "url": "https://news.ycombinator.com/item?id=47929130", "by": "simonw", "date": "2026-04-27", "version": null, "stance": "neutral", "quote": "Talkie isn't entirely pure either though", "excerpt": "I remain hopeful that some day someone will train an LLM which is tolerable to people who take this stance (which I respect, much like I respect food vegetarians despite not being one myself). I've been tracking models trained entirely on out-of-copyright data, for example. I've not yet seen one of those which appears generally useful and didn't chuck in a scrape of the web or get fine-tuned on examples generated by ", "context": "", "use_case": "", "switched_from": "", "story": "CS Professor: To My Students"}, {"id": 752706144, "item_id": 47044134, "url": "https://news.ycombinator.com/item?id=47044134", "by": "arach", "date": "2026-02-16", "version": null, "stance": "recommend", "quote": "Parakeet and incredibly fast and made for devs", "excerpt": "https://usetalkie.com - Parakeet and incredibly fast and made for devs", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Free Alternative to Wispr Flow, Superwhisper, and Monologue"}]}