[{"data":1,"prerenderedAt":329},["ShallowReactive",2],{"project-49696402":3},{"id":4,"name":5,"line":5,"line_source":6,"title":7,"url":8,"domain":9,"maker":10,"posted_at":11,"year":12,"points":13,"comments":14,"rank_in_year":15,"shown_in_year":16,"top":17,"tier":18,"category":19,"subcategory":20,"format":21,"keywords":22,"reception":27,"sentiment":28,"stance_mix":29,"scored_comments":34,"aspects":35,"strip":43,"post_text":44,"what":45,"said":60,"maker_in_thread":156,"shape":158,"grid":163,"attempts":290,"same_week":291,"maker_shows":160,"maker_hidden":321,"reader_version":322,"card":323},"49696402","Pelican-bicycle alternatives","title","Show HN: Pelican-bicycle alternatives","https:\u002F\u002Fgally.net\u002Ftemp\u002F20260914pelican-alternatives\u002Findex.html","gally.net","tkgally",1789392022,2026,133,45,494,4366,12,"notable","ai","model benchmark","web_app",[23,24,25,26],"llm benchmark","svg generation","model comparison","pelican riding bicycle","mixed",-0.032,{"positive":30,"neutral":31,"mixed":32,"critical":33},8,13,3,10,34,{"idea":36,"execution":38,"design":41},{"positive":37,"critical":32},1,{"positive":39,"critical":40},5,7,{"positive":37,"critical":42},0,[32,37,42,42,42,32,37,37,32,42,32,32,42,32,32,32,39,37,32,32,42,42,37,39,32,37,42,42,42,39,32,37,39,32,32,39,32,39,39,37],"In November and December 2025, inspired by Simon Willison’s pelican-riding-a-bicycle benchmark, I had some then-current LLMs create SVGs from thirty similar prompts, such as “Generate an SVG of an octopus operating a pipe organ.” Simon mentioned that experiment on his blog [1].\n\nNine months have passed and much stronger models have been released, so I tried the experiment again today. The linked site shows the results.\n\nRunning ten of the prompts through six models at OpenRouter cost about twenty dollars, so I stopped there for now.\n\n[1] https:\u002F\u002Fsimonwillison.net\u002F2025\u002FNov\u002F25\u002F",{"problem":46,"audience":52,"pricing":52,"motivation":53,"time_to_build":52,"stack":57,"openness":58,"stage":59,"team_size":58},{"value":47,"source":48,"evidence":49},"Compares how well LLMs generate SVG drawings from whimsical prompts","maker",[50],{"id":51},"post",null,{"value":54,"source":48,"evidence":55},"Retest the pelican-on-bicycle SVG experiment with newer models nine months later",[56],{"id":51},[],"unknown","launched",{"praise":61,"objections":84,"requests":115,"questions":134,"alternatives":141},[62,70],{"value":63,"source":64,"evidence":65},"Visual comparison is easier to grasp than numeric scores","commenter",[66],{"id":67,"by":68,"text":69},"49703275","simonw","Because it's one of the few ways of comparing models that lets you instantly evaluate them visually. That makes it more comprehensible than a numeric score on a benchmark.",{"value":71,"source":64,"evidence":72},"Some models' SVG outputs are impressive and cute",[73,77,80],{"id":74,"by":75,"text":76},"49703374","water-drummer","Ok the Grok ones are cute",{"id":78,"by":68,"text":79},"49703246","I love these.",{"id":81,"by":82,"text":83},"49709063","theshrike79","What's interesting to me is that the SVG versions don't have the \"AI Image generation hates negative space\" issue as badly as generated images do. They kinda stay on point and…",[85,100],{"value":86,"source":64,"evidence":87},"Benchmark appears Goodharted or saturated by training on similar SVG tasks",[88,92,96],{"id":89,"by":90,"text":91},"49698369","vova_hn2","Website looks very cool, Fable's octopus-organist looks very cute, but I feel like this benchmark (generate an SVG by a short and slightly ridiculous description) in general has…",{"id":93,"by":94,"text":95},"49699176","neilellis","Well that benchmark is now saturated, what next. How fast you can hack the pentagon?",{"id":97,"by":98,"text":99},"49703359","gpt5","It’s very likely they all add svg generation into the training data. It’s part of the reason it’s no longer a good benchmark (unless you need to generate SVGs).",{"value":101,"source":64,"evidence":102},"Outputs often incoherent, e.g. Gemini Flash results and zebras",[103,107,111],{"id":104,"by":105,"text":106},"49699360","GaggiX","I don't think a bunch of similar tasks can really saturate the \"create a SVG of X\", because the model should have a quite good spatial understanding of the world and how…",{"id":108,"by":109,"text":110},"49699535","Jordan-117","Better than Astra and Fable? It looks quite pretty and even impressive at times if you squint, but look closer and it falls apart in terms of coherency. And I say that as…",{"id":112,"by":113,"text":114},"49699015","CamperBob2","All models are pretty good now at generating these images. Not zebras. If you want to see how bad SVG output still is, ask for a zebra riding a scooter.",[116,127],{"value":117,"source":64,"evidence":118},"Add Qwen small local models such as Qwen 3.8 Flash-Next",[119,123],{"id":120,"by":121,"text":122},"49700912","kennywinker","Would love to see Qwen3.8-27b here, since that is the model most people are running locally.",{"id":124,"by":125,"text":126},"49712588","meta-level","Qwen 3.8 Flash-Next is really missing here (because it's small enough to run it locally on a (formerly) affordable machine). This is what Qwen3.8-Flash-Next-UD-Q4_K_XL gives me…",{"value":128,"source":64,"evidence":129},"Add iterative render-and-fix runs for cheaper models",[130],{"id":131,"by":132,"text":133},"49704827","miohtama","I hope there would be parameter iterations allowing especially cheaper models on inspect and fix their output via rendering",[135],{"value":136,"source":64,"evidence":137},"Why is this a good test?",[138],{"id":139,"by":140,"text":136},"49700020","dcreater",[142,149],{"value":143,"source":64,"evidence":144},"Little Dorrit Benchmark for visual reasoning and structured output",[145],{"id":146,"by":147,"text":148},"49698900","svcrunch","I'd like to mention the Little Dorrit Benchmark [1] which I have been running for a couple of years now. It has a few nice features: 1. It tests visual reasoning and structured…",{"value":150,"source":64,"evidence":151},"MacBook Pro 3D SVG benchmark from playcode.io",[152],{"id":153,"by":154,"text":155},"49703545","ianberdin","https:\u002F\u002Fplaycode.io\u002Fblog\u002Fmacbook-svg-benchmark MacBook Pro 3D in SVG for me the most helpful one.",{"comments":42,"answered":42,"key":157},[],{"commenters":159,"max_depth":160,"maker_replies":42,"first_comment_mins":161,"lifespan_hours":162},40,4,79,25,[164,166,168,170,174,177,182,184,187,190,194,197,200,201,202,205,208,212,213,216,219,222,223,226,229,233,237,241,243,246,248,249,251,255,258,260,262,264,268,272,275,278,280,283,287],[146,42,32,165,147,42],"project\u002Fneutral\u002Fnone",[89,37,37,167,90,42],"project\u002Fcritical\u002Fidea",[104,37,42,169,105,42],"commenter\u002Fneutral\u002Fnone",[171,37,42,172,173,42],"49702321","other\u002Fneutral\u002Fnone","andy_ppp",[175,37,42,172,176,42],"49707570","pampas",[178,179,32,180,181,42],"49697951",2,"project\u002Fmixed\u002Fexecution","samayashar",[112,179,37,183,113,42],"project\u002Fcritical\u002Fexecution",[185,179,37,183,186,42],"49704283","rukuu001",[188,179,37,183,189,42],"49699562","honeycrispy",[191,32,32,192,193,42],"49701204","project\u002Fneutral\u002Fidea","outlore",[195,32,42,169,196,42],"49701240","Terr_",[198,32,32,165,199,42],"49705387","trebligdivad",[120,160,32,165,121,42],[153,39,42,172,154,42],[124,203,32,204,125,42],6,"project\u002Fneutral\u002Fexecution",[206,40,32,165,207,42],"49697663","steinvakt2",[209,30,39,210,211,42],"49698371","project\u002Fpositive\u002Fexecution","sajithdilshan",[108,30,37,183,109,42],[81,214,39,215,82,42],9,"project\u002Fpositive\u002Fdesign",[217,33,37,183,218,42],"49700443","zaphar",[220,221,32,180,173,42],"49702280",11,[131,17,32,192,132,42],[224,31,42,172,225,42],"49700447","mock-possum",[227,31,42,172,228,42],"49702584","amysox",[230,231,37,183,232,42],"49698486",14,"sceptic123",[234,235,32,192,236,42],"49699046",15,"eddytrex_",[238,239,39,210,240,42],"49699182",16,"BrokenCogs",[242,239,32,180,105,42],"49699269",[244,239,37,183,245,42],"49700404","pixelesque",[247,239,42,172,68,42],"49703267",[97,239,42,169,98,42],[250,239,42,169,68,42],"49703457",[252,253,39,210,254,42],"49703314",17,"pohl",[256,253,32,165,257,42],"49703572","eleventen",[93,259,37,167,94,42],18,[139,261,32,192,140,42],19,[67,261,39,263,68,42],"maker\u002Fpositive\u002Fidea",[265,266,32,192,267,42],"49699919",20,"dustfinger",[269,270,32,192,271,42],"49698781",21,"qiine",[273,270,39,210,274,42],"49699990","ormax3",[276,270,32,204,277,42],"49700497","input_sh",[74,279,39,210,75,42],22,[78,281,39,282,68,42],23,"project\u002Fpositive\u002Fnone",[284,285,37,167,286,42],"49699717",24,"villish",[288,285,42,169,289,42],"49708495","bahmboo",[],[292,297,302,307,312,316],{"id":293,"name":294,"points":295,"category":296},"49711544","Fugleramme",2391,"creative",{"id":298,"name":299,"points":300,"category":301},"49637385","rivendell.dmitrybrant.com",633,"learning_science",{"id":303,"name":304,"points":305,"category":306},"49693971","Inflight Simulator",446,"games",{"id":308,"name":309,"points":310,"category":311},"49712278","Capsule",380,"dev_tools",{"id":313,"name":314,"points":315,"category":301},"49649789","Bodily Oddities",341,{"id":317,"name":318,"points":319,"category":320},"49712102","modem-thing",210,"hardware",false,"v2",{"id":324,"name":5,"line":5,"ai":321,"cat":19,"year":12,"points":13,"comments":14,"rec":27,"s":325,"top":17,"kw":326,"maker":10,"domain":9,"tier":18,"attempts":42,"strip":327,"url":8,"mix":328,"praise":63,"objection":86,"sub":20},49696402,-0.03,[23,24,25,26],"3100031130330333513300153100053153353551",[30,31,32,33],1794036710050]