{"slug": "functiongemma", "name": "FunctionGemma", "category": "ai-models", "category_name": "AI models", "description": "", "recommenders": 4, "commenters": 5, "warners": 1, "score": 2.472, "rising": 0.02, "first": "2025-12-18", "latest": "2026-09-04", "why": "Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 4, "neutral": 0, "compare": 0, "warn": 1}, "spark": [0, 2, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy.", "kind": "", "icon": true, "family": "gemma", "domain": "blog.google", "github_repo": null, "official_url": "https://ai.google.dev/gemma/docs/functiongemma", "pricing": "", "aliases": ["FunctionGemma"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 793099472, "item_id": 49568717, "url": "https://news.ycombinator.com/item?id=49568717", "by": "kouteiheika", "date": "2026-09-04", "version": null, "stance": "recommend", "quote": "Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy.", "excerpt": "> Models like ornith:9b, mistral:7b or cogito:14b can get the job done sometimes, but they are not fast and reliable enough for general use, specially if you have only 4GB of VRAM. Have you considered/tried using a model that's, well, more appropriate size-wise for an use case like this? These are relatively big. Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy. [1] https", "context": "", "use_case": "small function-calling model", "switched_from": "", "story": "Show HN: TERMy \u2013 A fast terminal assistant that does not use LLMs"}, {"id": 747195140, "item_id": 46699696, "url": "https://news.ycombinator.com/item?id=46699696", "by": "jauntywundrkind", "date": "2026-01-20", "version": null, "stance": "recommend", "quote": "Tiny models like FunctionGemma will be able to take simple commands and get the work done", "excerpt": "What had Qwen3-coder rejected? Right now that seems like the strongest recommendation. GLM-4.7-Flash seems very promising but is so new. Gemma3 is also very good. Nanbeige-4 is supposedly incredibly capable. Both are very small. https://huggingface.co/google/gemma-3-4b-it https://huggingface.co/Nanbeige/Nanbeige4-3B-Thinking-2511 Ideally IMO, you should probably build little tools or a multi-tool for doing the work y", "context": "", "use_case": "", "switched_from": "", "story": "Local models to support home network infrastructure?"}, {"id": 741113040, "item_id": 46319565, "url": "https://news.ycombinator.com/item?id=46319565", "by": "canyon289", "date": "2025-12-18", "version": null, "stance": "recommend", "quote": "FunctionGemma can work just like what's intended here", "excerpt": "I've only just skimmed this blog post but if I'm reading correctly FunctionGemma can work just like what's intended here, a \"contextless\" tool router. Going one level up you as a developer have a choice how much context you want to provide to the model. Philipp Schmid wrote a good blog post about this, titling this \"context engineering\". I like his idea because instead of just blindly throwing stuff into a model's co", "context": "", "use_case": "On-device tool routing", "switched_from": "", "story": "FunctionGemma 270M Model"}, {"id": 741074928, "item_id": 46317183, "url": "https://news.ycombinator.com/item?id=46317183", "by": "xnx", "date": "2025-12-18", "version": null, "stance": "recommend", "quote": "Amazing it can run in the browser.", "excerpt": "Cool game! Amazing it can run in the browser. My mind was blown when I saw you could give goal based commands vs prescriptive ones. https://huggingface.co/spaces/webml-community/FunctionGemma-Physics-Playground", "context": "", "use_case": "goal-based function calling demo", "switched_from": "", "story": "FunctionGemma 270M Model"}]}], "caveats": [{"label": "", "summary": "", "quotes": [{"id": 793103824, "item_id": 49568989, "url": "https://news.ycombinator.com/item?id=49568989", "by": "coder543", "date": "2026-09-04", "version": null, "stance": "warn", "quote": "FunctionGemma never worked well for me (without fine tuning)", "excerpt": "FunctionGemma never worked well for me (without fine tuning). Liquid has released 230M and 350M models that work far, far better in my testing: https://huggingface.co/LiquidAI/LFM2.5-230M I really look forward to a hypothetical LFM3-230M, because LFM2.5-230M is so close to being usable, while FunctionGemma is miles away from being usable. But, yes, still tangential to TERMy.", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: TERMy \u2013 A fast terminal assistant that does not use LLMs"}]}], "who": [], "tags": [], "use_cases": ["small function-calling model", "on-device tool routing", "goal-based function calling demo"], "compared": [], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=49562219", "title": "Show HN: TERMy \u2013 A fast terminal assistant that does not use LLMs", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=46316533", "title": "FunctionGemma 270M Model", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=46690846", "title": "Local models to support home network infrastructure?", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {"recommend": 2}, "2026-01": {"recommend": 1}, "2026-02": {}, "2026-03": {}, "2026-04": {}, "2026-05": {}, "2026-06": {}, "2026-07": {}, "2026-08": {}, "2026-09": {"recommend": 1, "warn": 1}, "2026-10": {}}, "versions": [], "version_unnamed": 5, "family_ref": {"slug": "gemma", "name": "Gemma"}, "runs_on": null, "lines": [], "quotes": [{"id": 793103824, "item_id": 49568989, "url": "https://news.ycombinator.com/item?id=49568989", "by": "coder543", "date": "2026-09-04", "version": null, "stance": "warn", "quote": "FunctionGemma never worked well for me (without fine tuning)", "excerpt": "FunctionGemma never worked well for me (without fine tuning). Liquid has released 230M and 350M models that work far, far better in my testing: https://huggingface.co/LiquidAI/LFM2.5-230M I really look forward to a hypothetical LFM3-230M, because LFM2.5-230M is so close to being usable, while FunctionGemma is miles away from being usable. But, yes, still tangential to TERMy.", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: TERMy \u2013 A fast terminal assistant that does not use LLMs"}, {"id": 793099472, "item_id": 49568717, "url": "https://news.ycombinator.com/item?id=49568717", "by": "kouteiheika", "date": "2026-09-04", "version": null, "stance": "recommend", "quote": "Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy.", "excerpt": "> Models like ornith:9b, mistral:7b or cogito:14b can get the job done sometimes, but they are not fast and reliable enough for general use, specially if you have only 4GB of VRAM. Have you considered/tried using a model that's, well, more appropriate size-wise for an use case like this? These are relatively big. Something like FunctionGemma [1] finetuned for a given set of tasks would be a lot more speedy. [1] https", "context": "", "use_case": "small function-calling model", "switched_from": "", "story": "Show HN: TERMy \u2013 A fast terminal assistant that does not use LLMs"}, {"id": 747195140, "item_id": 46699696, "url": "https://news.ycombinator.com/item?id=46699696", "by": "jauntywundrkind", "date": "2026-01-20", "version": null, "stance": "recommend", "quote": "Tiny models like FunctionGemma will be able to take simple commands and get the work done", "excerpt": "What had Qwen3-coder rejected? Right now that seems like the strongest recommendation. GLM-4.7-Flash seems very promising but is so new. Gemma3 is also very good. Nanbeige-4 is supposedly incredibly capable. Both are very small. https://huggingface.co/google/gemma-3-4b-it https://huggingface.co/Nanbeige/Nanbeige4-3B-Thinking-2511 Ideally IMO, you should probably build little tools or a multi-tool for doing the work y", "context": "", "use_case": "", "switched_from": "", "story": "Local models to support home network infrastructure?"}, {"id": 741113040, "item_id": 46319565, "url": "https://news.ycombinator.com/item?id=46319565", "by": "canyon289", "date": "2025-12-18", "version": null, "stance": "recommend", "quote": "FunctionGemma can work just like what's intended here", "excerpt": "I've only just skimmed this blog post but if I'm reading correctly FunctionGemma can work just like what's intended here, a \"contextless\" tool router. Going one level up you as a developer have a choice how much context you want to provide to the model. Philipp Schmid wrote a good blog post about this, titling this \"context engineering\". I like his idea because instead of just blindly throwing stuff into a model's co", "context": "", "use_case": "On-device tool routing", "switched_from": "", "story": "FunctionGemma 270M Model"}, {"id": 741074928, "item_id": 46317183, "url": "https://news.ycombinator.com/item?id=46317183", "by": "xnx", "date": "2025-12-18", "version": null, "stance": "recommend", "quote": "Amazing it can run in the browser.", "excerpt": "Cool game! Amazing it can run in the browser. My mind was blown when I saw you could give goal based commands vs prescriptive ones. https://huggingface.co/spaces/webml-community/FunctionGemma-Physics-Playground", "context": "", "use_case": "goal-based function calling demo", "switched_from": "", "story": "FunctionGemma 270M Model"}]}