[{"data":1,"prerenderedAt":355},["ShallowReactive",2],{"project-49471407":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":48,"post_text":50,"what":51,"said":77,"maker_in_thread":178,"shape":190,"grid":194,"attempts":317,"same_week":318,"maker_shows":42,"maker_hidden":347,"reader_version":348,"card":349},"49471407","Experiential","Open OpenRouter that turns usage into a better model","title","Show HN: We built open OpenRouter that turns usage into a better model","https:\u002F\u002Fgithub.com\u002Fexperientiallabs\u002Fexperiential","github.com","SilenN",1787865515,2026,222,47,241,4366,6,"notable","dev_tools","model gateway","api",[24,25,26,27],"llm gateway","model routing","openrouter alternative","llm proxy","curious",0.006,{"positive":31,"neutral":32,"mixed":33,"critical":34},7,17,0,9,33,{"performance":37,"idea":38,"docs":41,"pricing":43,"execution":45,"privacy":46,"design":47},{"positive":33,"critical":33},{"positive":39,"critical":40},4,5,{"positive":33,"critical":42},2,{"positive":44,"critical":33},1,{"positive":44,"critical":33},{"positive":33,"critical":42},{"positive":44,"critical":33},[49,49,33,44,33,49,49,33,40,33,44,44,40,33,44,40,49,49,33,49,49,33,44,40,44,33,49,49,49,44,33,44,33,49,49,49,40,49,40,40],3,"Hi HN, we built an open source model gateway. It's a single place to manage our own self hosted, frontier, and open source models in one place.\n\nIt’s is rust native, built for concurrency, and implements all the config quirks across models and providers (streaming formats, tool calls, model parameters, rate limits, and different error behavior).\n\nThe gateway adds under 1 ms for BYOK requests and under 2 ms when Experiential supplies the provider key. It has every major inference provider, and 1000+ models refreshed daily via a codex agent that opens a PR.\n\nCompared to other similar projects we",{"problem":52,"audience":58,"pricing":59,"motivation":58,"time_to_build":58,"stack":66,"openness":74,"stage":75,"team_size":76},{"value":53,"source":54,"evidence":55},"Managing many model providers and choosing the right model per request, with overhead and markup","maker",[56],{"id":57},"post",null,{"value":60,"source":54,"evidence":61},"No markup on tokens; enterprise licenses and hosting make money",[62,63],{"id":57},{"id":64,"by":11,"text":65},"49475516","Telemetry is off by default. PostHog is for usage analytics on the open source repo. Audit it if you're skeptical. We make money off enterprise licenses and hosting models. I…",[67],{"value":68,"source":54,"evidence":69},"Rust",[70,71],{"id":57},{"id":72,"by":11,"text":73},"49472292","Yep! If you look at the commit history that's exactly what happened.","open_source","launched","team",{"praise":78,"objections":94,"requests":117,"questions":130,"alternatives":145},[79,87],{"value":80,"source":81,"evidence":82},"Open source and no markup is the right default for a gateway","commenter",[83],{"id":84,"by":85,"text":86},"49473195","akshay_akula","Open source and no markup is the right default for a gateway. The caching question above is the one I would want answered before swapping models though.",{"value":88,"source":81,"evidence":89},"Router and model optimization from traffic is a useful differentiator",[90],{"id":91,"by":92,"text":93},"49472854","kfallah15","Router and model optimization from traffic is the main differentiator",[95,103,110],{"value":96,"source":81,"evidence":97},"Switching models may break cached input token savings",[98,102],{"id":99,"by":100,"text":101},"49471977","Areibman","Could you say more about how caching works? One major advantage of sticking with a single model is saving money on cached input tokens. I'd imagine if you swap between a bunch of…",{"id":84,"by":85,"text":86},{"value":104,"source":81,"evidence":105},"Telemetry is on by default despite open source claim",[106],{"id":107,"by":108,"text":109},"49478494","aHumbleUser","Your GitHub readme says telemetry is on by default",{"value":111,"source":81,"evidence":112},"Small team could reproduce it quickly",[113],{"id":114,"by":115,"text":116},"49475346","rdslw","impressive only if using pre-gpt era assumptions about saas\u002Fproducts\u002Fsoftware. unfortunately a small team can reproduce it in two months, which greatly lowers value of it. we, as…",[118,123],{"value":119,"source":81,"evidence":120},"Explain how caching works",[121,122],{"id":99,"by":100,"text":101},{"id":84,"by":85,"text":86},{"value":124,"source":81,"evidence":125},"Support semantic caching at the router level",[126],{"id":127,"by":128,"text":129},"49474132","sangwook","What online signal recalibrates simulated rankings against actual task success? Also do you have a plan to support semantic caching at the router level?",[131,138],{"value":132,"source":81,"evidence":133},"How does the opt-in model training from traffic work",[134],{"id":135,"by":136,"text":137},"49476194","d2p","> and use your traffic to (opt in) train you a model. Is there more info on this? I'm curious exactly what it is. Is it fine-tuning\u002FLoRA on some base model? Don't cloud providers…",{"value":139,"source":81,"evidence":140},"Why text world models",[141],{"id":142,"by":143,"text":144},"49482736","carloslfu","why text world models? Is this backed by any evidence or did you find it to be good in practice?",[146,157,161,165,172],{"value":147,"source":81,"evidence":148},"LiteLLM",[149,153],{"id":150,"by":151,"text":152},"49477102","mongrelion","Congrats on the launching of your product. I will be taking it for a spin to compare it with these other products that seem to be competing directly with what you have to offer:…",{"id":154,"by":155,"text":156},"49472738","cheema33","I have not tried it yet. Is it similar to LiteLLM? If so, what sets it apart?",{"value":158,"source":81,"evidence":159},"Bifrost",[160],{"id":150,"by":151,"text":152},{"value":162,"source":81,"evidence":163},"GoModel",[164],{"id":150,"by":151,"text":152},{"value":166,"source":81,"evidence":167},"vLLM Semantic Router",[168],{"id":169,"by":170,"text":171},"49476057","nejch","A part of me wishes the open source community would focus making research and industry-backed initiatives like the vLLM Semantic Router rock solid. Then I'd spend less time every…",{"value":173,"source":81,"evidence":174},"ngrok AI gateway",[175],{"id":176,"by":85,"text":177},"49478066","You guys should look into ngrok ai gateway. We have some small models running on local hardware that we tried to use but it was too painful. Just wanted to not waste all the…",{"comments":179,"answered":40,"key":180},12,[181,184,187],{"id":182,"text":183},"49472077","The trick is to rarely switch, or switch at task boundaries. Often the conclusion of routing is actually \"this one model is actually at the pareto front for this task, just use…",{"id":185,"text":186},"49473093","That is another way to do. Or we can automatically figure out which models the subagents should be using for you. And update them as new models come out and the work your…",{"id":188,"text":189},"49479962","We had this exact problem so we solved it for ourselves. Happy to help if you run into any issues. I have a \u002Fhmmm command I use for the second part that works reasonably well:…",{"commenters":191,"max_depth":40,"maker_replies":179,"first_comment_mins":192,"lifespan_hours":193},30,54,85,[195,197,200,201,205,207,211,215,216,220,224,226,229,232,233,235,237,238,241,243,244,245,247,251,255,256,260,263,267,268,271,274,278,281,283,286,287,288,290,294,297,299,301,305,307,310,315],[99,33,49,196,100,33],"project\u002Fneutral\u002Fperformance",[198,33,49,196,199,33],"49472056","purplecats",[182,33,33,196,11,44],[202,33,44,203,204,33],"49473066","project\u002Fcritical\u002Fidea","cameronh90",[185,33,33,206,11,44],"maker\u002Fneutral\u002Fidea",[208,33,33,209,210,33],"49476766","commenter\u002Fneutral\u002Fnone","aerzen",[212,33,49,213,214,33],"49474815","project\u002Fneutral\u002Fidea","try-working",[150,44,49,213,151,33],[217,44,33,218,219,33],"49508063","commenter\u002Fneutral\u002Fperformance","santiago-pl",[221,42,40,222,223,33],"49478084","project\u002Fpositive\u002Fidea","jakswa",[188,42,33,225,11,44],"maker\u002Fpositive\u002Fnone",[227,42,44,228,223,33],"49483508","project\u002Fcritical\u002Fdocs",[230,49,44,228,231,33],"49476304","croemer",[169,39,44,203,170,33],[84,40,40,234,85,33],"project\u002Fpositive\u002Fpricing",[236,40,33,213,11,44],"49473330",[176,40,44,203,85,33],[239,18,40,222,240,33],"49473165","ceroxylon",[242,18,49,213,92,33],"49473292",[154,31,49,213,155,33],[91,31,49,213,92,33],[246,31,33,213,11,44],"49473103",[248,249,49,213,250,33],"49478018",8,"naw103",[252,34,49,253,254,33],"49472268","project\u002Fneutral\u002Fexecution","23david",[72,34,33,253,11,44],[257,258,44,203,259,33],"49473410",10,"0xbadcafebee",[261,258,33,262,11,44],"49473543","maker\u002Fneutral\u002Fnone",[264,258,40,265,266,33],"49474662","project\u002Fpositive\u002Fexecution","tyre",[114,258,44,203,115,33],[269,258,33,270,11,44],"49475541","maker\u002Fneutral\u002Fexecution",[135,272,49,273,136,33],11,"project\u002Fneutral\u002Fprivacy",[275,179,49,276,277,33],"49474138","project\u002Fneutral\u002Fpricing","forgetme2020",[279,179,49,276,280,33],"49474187","kakugawa",[282,179,49,276,92,33],"49474250",[284,179,44,285,115,33],"49475391","project\u002Fcritical\u002Fprivacy",[64,179,33,273,11,44],[107,179,44,285,108,33],[289,179,33,273,11,44],"49480973",[291,292,49,213,293,33],"49473874",13,"swthbht",[295,292,33,296,11,44],"49473983","project\u002Fpositive\u002Fperformance",[127,298,49,196,128,33],14,[300,298,49,213,92,33],"49474234",[302,303,40,222,304,33],"49484242",15,"bicepjai",[142,306,49,213,143,33],16,[308,32,40,222,309,33],"49471998","ashermania",[311,312,40,313,314,33],"49475767",18,"project\u002Fpositive\u002Fdesign","gpiechnik2",[316,312,33,262,11,44],"49476010",[],[319,324,329,333,338,342],{"id":320,"name":321,"points":322,"category":323},"49461817","load-bearing",710,"ai",{"id":325,"name":326,"points":327,"category":328},"49489376","Laser Graffiti",253,"creative",{"id":330,"name":331,"points":332,"category":323},"49524447","slotstream",240,{"id":334,"name":335,"points":336,"category":337},"49471965","OpenTIE",231,"games",{"id":339,"name":340,"points":336,"category":341},"49478426","wilsonharper.net","hardware",{"id":343,"name":344,"points":345,"category":346},"49528895","Weedout",185,"consumer_social",false,"v2",{"id":350,"name":5,"line":6,"ai":347,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":351,"top":18,"kw":352,"maker":11,"domain":10,"tier":19,"attempts":33,"strip":353,"url":9,"mix":354,"praise":80,"objection":96,"sub":21},49471407,0.01,[24,25,26,27],"3301033050115015330330151033310103335355",[31,32,33,34],1794036708916]