[{"data":1,"prerenderedAt":134},["ShallowReactive",2],{"project-44768159":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"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":28,"sentiment":9,"stance_mix":29,"scored_comments":14,"aspects":32,"strip":34,"post_text":37,"what":38,"said":61,"maker_in_thread":87,"shape":89,"grid":91,"attempts":96,"same_week":97,"maker_shows":35,"maker_hidden":126,"reader_version":127,"card":128},"44768159","Predict","Predict GPT-5 skills with a community AI benchmark","title","Show HN: Predict GPT-5 skills with a community AI benchmark",null,"andrewxhill",1754146833,2025,13,2,2487,3837,65,"shown","ai","ai benchmarking","web_app",[23,24,25,26,27],"ai benchmark","llm evaluation","model forecasting","gpt-5","community evals","quiet",{"positive":30,"neutral":30,"mixed":31,"critical":31},1,0,{"idea":33},{"positive":30,"critical":31},[35,36],3,5,"I’m Andrew, co-founder of Recall. Over the past few days I’ve been building Predict, a playground where anyone can:\n\n- propose skills we should measure in language models—live examples include difficult math, memory-manipulation resistance, code generation, and empathy under bad news\n\n- write evals (graded prompts) for those skills\n\n- forecast which models will score highest once GPT-5 is released\n\nWhy this exists\n\nBenchmarks leak into training data quickly; scores are unreliable and labs still declare progress. The prediction tool aims keeps the target moving by letting the crowd define both ",{"problem":39,"audience":45,"pricing":9,"motivation":49,"time_to_build":9,"stack":53,"openness":58,"stage":59,"team_size":60},{"value":40,"source":41,"evidence":42},"Public benchmarks leak into training data and scores are unreliable","maker",[43],{"id":44},"post",{"value":46,"source":41,"evidence":47},"Anyone interested in predicting and evaluating language model skills",[48],{"id":44},{"value":50,"source":41,"evidence":51},"Keep the benchmark target moving by letting the crowd define tasks and scoring",[52],{"id":44},[54],{"value":55,"source":41,"evidence":56},"Built almost entirely with Claude Code",[57],{"id":44},"unknown","beta","team",{"praise":62,"objections":75,"requests":76,"questions":81,"alternatives":86},[63],{"value":64,"source":65,"evidence":66},"Commenters find the product interesting and show early traction","commenter",[67,71],{"id":68,"by":69,"text":70},"44768209","s3n4","Congratulations on the experiment. Super interesting product, and seems like a lot of traction already with more than 700K performance predictions. How will the system itself…",{"id":72,"by":73,"text":74},"44779026","vondragon","Love this. We need this. Would love to help out and integrate with modelz.io",[],[77],{"value":78,"source":65,"evidence":79},"Explain how submitted skills and evals are vetted and included",[80],{"id":68,"by":69,"text":70},[82],{"value":83,"source":65,"evidence":84},"How does the system evaluate community skills and evals before including them?",[85],{"id":68,"by":69,"text":70},[],{"comments":31,"answered":31,"key":88},[],{"commenters":14,"max_depth":30,"maker_replies":31,"first_comment_mins":36,"lifespan_hours":90},28,[92,94],[68,31,35,93,69,31],"project\u002Fneutral\u002Fidea",[72,30,36,95,73,31],"project\u002Fpositive\u002Fidea",[],[98,103,107,112,117,121],{"id":99,"name":100,"points":101,"category":102},"44787902","Kilopixel",1145,"creative",{"id":104,"name":105,"points":106,"category":19},"44807868","Kitten TTS",1003,{"id":108,"name":109,"points":110,"category":111},"44792005","Carbon",322,"web_saas",{"id":113,"name":114,"points":115,"category":116},"44747229","Mcp-use",155,"dev_tools",{"id":118,"name":119,"points":120,"category":19},"44767775","browser-llm",145,{"id":122,"name":123,"points":124,"category":125},"44801399","Whittle",141,"games",false,"v2",{"id":129,"name":5,"line":6,"ai":126,"cat":19,"year":12,"points":13,"comments":14,"rec":28,"s":9,"top":17,"kw":130,"maker":10,"domain":131,"tier":18,"attempts":31,"strip":132,"url":-1,"mix":133,"praise":64,"objection":-1,"sub":20},44768159,[23,24,25,26,27],"","35",[30,30,31,31],1794036679504]