{"slug": "gliner", "name": "GLiNER", "category": "ai", "category_name": "AI models and assistants", "description": "", "recommenders": 5, "commenters": 9, "warners": 1, "score": 4.394, "rising": 1.02, "first": "2026-03-03", "latest": "2026-10-27", "why": "Also, Gliner or similar models for different uses.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 5, "neutral": 2, "compare": 1, "warn": 1}, "spark": [0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 1, 1], "listed": true, "loved": [], "caveat": "", "short_quote": "Also, Gliner or similar models for different uses.", "kind": "", "icon": false, "domain": null, "github_repo": "github.com/urchade/gliner", "official_url": "https://github.com/urchade/GLiNER", "pricing": "", "aliases": ["GLiNER"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 804240160, "item_id": 50265010, "url": "https://news.ycombinator.com/item?id=50265010", "by": "samstevens", "date": "2026-10-27", "stance": "recommend", "quote": "gliner-style models are absolutely crazy", "excerpt": "gliner-style models are absolutely crazy, I learned about them a couple weeks ago when a co-worker fine-tuned an LFM encoder for promptable classification. as far as I know, you don't need to fine-tune; you can just prompt for classification. My co-worker's demo is here: https://huggingface.co/spaces/LiquidAI/prompt-routing", "context": "", "use_case": "", "switched_from": "", "story": "Introducing LFM3: Every Modality for Specialized Intelligence"}, {"id": 796263920, "item_id": 49766495, "url": "https://news.ycombinator.com/item?id=49766495", "by": "kamranjon", "date": "2026-09-19", "stance": "recommend", "quote": "It is also open source[3] and I think also has some good usage.", "excerpt": "It is really interesting to see this claim, because i thought the current theory was that typesafe actually repackaged the work from GLiNER[1] - which does seem to be a closer match, and their original paper[2] predates yours by several years. Curious if you had heard of it before? It is also open source[3] and I think also has some good usage. [1] https://arxiv.org/abs/2507.18546 [2] https://arxiv.org/abs/2311.08526", "context": "", "use_case": "schema-guided extraction", "switched_from": "", "story": "I built non-autoregressive decision models with RL a year ago"}, {"id": 781322548, "item_id": 48832659, "url": "https://news.ycombinator.com/item?id=48832659", "by": "brainless", "date": "2026-07-08", "stance": "recommend", "quote": "Also, Gliner or similar models for different uses.", "excerpt": "I have a project on very similar lines, https://github.com/brainless/dwata, which I have not been developing for the past few months. I have been meaning to get back to it and I really like what I see on your project page. My aim is to build a truly local app using only tiny/small models. I have had really good results from Qwen 3.5 1B, 4B, etc. Also, Gliner or similar models for different uses. SQLite + sqlite-vec +", "context": "", "use_case": "entity extraction", "switched_from": "", "story": "Show HN: Rowboat \u2013 Open-source, local-first alternative to Claude Desktop"}, {"id": 778001888, "item_id": 48625118, "url": "https://news.ycombinator.com/item?id=48625118", "by": "deepsquirrelnet", "date": "2026-06-21", "stance": "recommend", "quote": "Zero-shot encoders like tasksource or GliNER", "excerpt": "If you want to go deeper on language models, try these project ideas: - Zero-shot encoders like tasksource or GliNER - Natural language inference: https://huggingface.co/blog/dleemiller/nli-xenc-ways-to-use - GRPO training - GEPA prompt tuning Qwen 0.6B (or GEPA, then GRPO) - Use an embedding model and train a classifier (MLP, logistic, svm) - Use a larger LLM to generate a synthetic dataset (beware of lack of divers", "context": "", "use_case": "", "switched_from": "", "story": "Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions"}, {"id": 756373344, "item_id": 47273334, "url": "https://news.ycombinator.com/item?id=47273334", "by": "fbilhaut", "date": "2026-03-06", "stance": "recommend", "quote": "GLiNER is a really great research work", "excerpt": "GLiNER is a really great research work. But putting this kind of things in production is just another job. Not trying to do self promotion here, but there are alternatives for this purpose, like gline-rs (https://github.com/fbilhaut/gline-rs). Support of GLiNER 2 models is on the way.", "context": "", "use_case": "Information extraction", "switched_from": "", "story": "GLiNER2: Unified Schema-Based Information Extraction"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 803777842, "item_id": 50236115, "url": "https://news.ycombinator.com/item?id=50236115", "by": "nico", "date": "2026-10-25", "stance": "warn", "quote": "At around 4 classes already LLM-based classifiers (Jev/Laya/GLiNER et al) start falling behind", "excerpt": "> and not using in anger Is this a typo? are you saying that this is so new people haven't struggled with Jev enough yet? very interesting > it starts to break down over 20 classifications Yes, I recently ran a benchmark on exactly that: https://jeffyclassify.com/blog/when-llms-fail-at-classification/ At around 4 classes already LLM-based classifiers (Jev/Laya/GLiNER et al) start falling behind", "context": "", "use_case": "", "switched_from": "", "story": "A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis"}]}], "who": [], "tags": [], "use_cases": ["schema-guided extraction", "entity extraction", "information extraction", "nlp-based pii detection"], "compared": [{"slug": "jev", "name": "Jev", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=47266736", "title": "GLiNER2: Unified Schema-Based Information Extraction", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=50262229", "title": "Introducing LFM3: Every Modality for Specialized Intelligence", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50233779", "title": "A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49765348", "title": "I built non-autoregressive decision models with RL a year ago", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49717558", "title": "Introducing System One Models and Jev", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=48819808", "title": "Show HN: Rowboat \u2013 Open-source, local-first alternative to Claude Desktop", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {"neutral": 2, "recommend": 1}, "2026-04": {}, "2026-05": {}, "2026-06": {"recommend": 1}, "2026-07": {"recommend": 1}, "2026-08": {}, "2026-09": {"compare": 1, "recommend": 1}, "2026-10": {"warn": 1, "recommend": 1}}, "quotes": [{"id": 804240160, "item_id": 50265010, "url": "https://news.ycombinator.com/item?id=50265010", "by": "samstevens", "date": "2026-10-27", "stance": "recommend", "quote": "gliner-style models are absolutely crazy", "excerpt": "gliner-style models are absolutely crazy, I learned about them a couple weeks ago when a co-worker fine-tuned an LFM encoder for promptable classification. as far as I know, you don't need to fine-tune; you can just prompt for classification. My co-worker's demo is here: https://huggingface.co/spaces/LiquidAI/prompt-routing", "context": "", "use_case": "", "switched_from": "", "story": "Introducing LFM3: Every Modality for Specialized Intelligence"}, {"id": 803777842, "item_id": 50236115, "url": "https://news.ycombinator.com/item?id=50236115", "by": "nico", "date": "2026-10-25", "stance": "warn", "quote": "At around 4 classes already LLM-based classifiers (Jev/Laya/GLiNER et al) start falling behind", "excerpt": "> and not using in anger Is this a typo? are you saying that this is so new people haven't struggled with Jev enough yet? very interesting > it starts to break down over 20 classifications Yes, I recently ran a benchmark on exactly that: https://jeffyclassify.com/blog/when-llms-fail-at-classification/ At around 4 classes already LLM-based classifiers (Jev/Laya/GLiNER et al) start falling behind", "context": "", "use_case": "", "switched_from": "", "story": "A 10-year-old CNN outperforms frontier LLMs for skin cancer diagnosis"}, {"id": 796263920, "item_id": 49766495, "url": "https://news.ycombinator.com/item?id=49766495", "by": "kamranjon", "date": "2026-09-19", "stance": "recommend", "quote": "It is also open source[3] and I think also has some good usage.", "excerpt": "It is really interesting to see this claim, because i thought the current theory was that typesafe actually repackaged the work from GLiNER[1] - which does seem to be a closer match, and their original paper[2] predates yours by several years. Curious if you had heard of it before? It is also open source[3] and I think also has some good usage. [1] https://arxiv.org/abs/2507.18546 [2] https://arxiv.org/abs/2311.08526", "context": "", "use_case": "schema-guided extraction", "switched_from": "", "story": "I built non-autoregressive decision models with RL a year ago"}, {"id": 795568049, "item_id": 49723003, "url": "https://news.ycombinator.com/item?id=49723003", "by": "flowerboy-t", "date": "2026-09-16", "stance": "compare", "quote": "do you all see the use cases being similar to what you might use Fastino's Gliner models for?", "excerpt": "do you all see the use cases being similar to what you might use Fastino's Gliner models for? i see similar differentiation from general purpose LLMs in the sense that they can take natural-language input and return outputs adherent to a user-defined schema. https://fastino.ai/blog/gliner2-5-span-free-information-extraction im thinking about how well Jev could be used to replace a current LLM-as-Judge evaluation work", "context": "", "use_case": "", "switched_from": "", "story": "Introducing System One Models and Jev"}, {"id": 781322548, "item_id": 48832659, "url": "https://news.ycombinator.com/item?id=48832659", "by": "brainless", "date": "2026-07-08", "stance": "recommend", "quote": "Also, Gliner or similar models for different uses.", "excerpt": "I have a project on very similar lines, https://github.com/brainless/dwata, which I have not been developing for the past few months. I have been meaning to get back to it and I really like what I see on your project page. My aim is to build a truly local app using only tiny/small models. I have had really good results from Qwen 3.5 1B, 4B, etc. Also, Gliner or similar models for different uses. SQLite + sqlite-vec +", "context": "", "use_case": "entity extraction", "switched_from": "", "story": "Show HN: Rowboat \u2013 Open-source, local-first alternative to Claude Desktop"}, {"id": 778001888, "item_id": 48625118, "url": "https://news.ycombinator.com/item?id=48625118", "by": "deepsquirrelnet", "date": "2026-06-21", "stance": "recommend", "quote": "Zero-shot encoders like tasksource or GliNER", "excerpt": "If you want to go deeper on language models, try these project ideas: - Zero-shot encoders like tasksource or GliNER - Natural language inference: https://huggingface.co/blog/dleemiller/nli-xenc-ways-to-use - GRPO training - GEPA prompt tuning Qwen 0.6B (or GEPA, then GRPO) - Use an embedding model and train a classifier (MLP, logistic, svm) - Use a larger LLM to generate a synthetic dataset (beware of lack of divers", "context": "", "use_case": "", "switched_from": "", "story": "Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions"}, {"id": 756488081, "item_id": 47280505, "url": "https://news.ycombinator.com/item?id=47280505", "by": "hbcondo714", "date": "2026-03-06", "stance": "neutral", "quote": "there is a dependency on gliner1", "excerpt": "Okay but there is a dependency on gliner1: https://github.com/fastino-ai/GLiNER2/issues/69", "context": "", "use_case": "", "switched_from": "", "story": "GLiNER2: Unified Schema-Based Information Extraction"}, {"id": 756373344, "item_id": 47273334, "url": "https://news.ycombinator.com/item?id=47273334", "by": "fbilhaut", "date": "2026-03-06", "stance": "recommend", "quote": "GLiNER is a really great research work", "excerpt": "GLiNER is a really great research work. But putting this kind of things in production is just another job. Not trying to do self promotion here, but there are alternatives for this purpose, like gline-rs (https://github.com/fbilhaut/gline-rs). Support of GLiNER 2 models is on the way.", "context": "", "use_case": "Information extraction", "switched_from": "", "story": "GLiNER2: Unified Schema-Based Information Extraction"}, {"id": 756288528, "item_id": 47268033, "url": "https://news.ycombinator.com/item?id=47268033", "by": "hbcondo714", "date": "2026-03-05", "stance": "neutral", "quote": "https://github.com/urchade/GLiNER", "excerpt": "I guess this is version 1, but still being maintained? https://github.com/urchade/GLiNER", "context": "", "use_case": "", "switched_from": "", "story": "GLiNER2: Unified Schema-Based Information Extraction"}, {"id": 755724626, "item_id": 47232789, "url": "https://news.ycombinator.com/item?id=47232789", "by": "mnagas", "date": "2026-03-03", "stance": "neutral", "quote": "Cloud mode adds NLP-powered detection (GLiNER) on top of the regex layer.", "excerpt": "We've been building AI apps for the past year and the PII problem kept nagging us. You want to use GPT/Claude for customer support, medical summaries, financial analysis \u2014 but you can't just ship user emails, SSNs, and health records to a third-party API. Compliance teams say no. Often, the law says no. The result: teams either avoid AI entirely or roll the dice with unprotected data. Both outcomes suck. We built Bli", "context": "", "use_case": "NLP-based PII detection", "switched_from": "", "story": "Show HN: Blindfold \u2013 PII protection for LLM apps (local regex and cloud NLP)"}]}