{"config":"agent_steerability","id":"arena-agent-steerability","title":"Agents · steerability","category":"Coding and agents","metric":"IPS score","direction":"higher","precision":6,"publishedAt":"2026-11-04","interval":"95% confidence interval","observationLabel":"Observations","description":"Agent Arena signal scores from the official dataset. Each signal is a separate measurement; it is not a success percentage.","limitations":["The population is the models in this source board, including variants and models without an available API price.","Ranks use the published confidence bounds: one plus the number of models whose lower bound exceeds this model’s upper bound. Shared ranks are not proof of equal performance."],"omissions":[],"rows":[{"name":"Gemini 4 Carbon (High)","organization":"google","score":0.1931931290535741,"low":0.18605803991074868,"high":0.20032821819639954,"observations":208968,"sourceRank":1,"publishedAt":"2026-11-04","rank":1},{"name":"Claude Opus 5.5 (Max)","organization":"anthropic","score":0.18428280355470372,"low":0.16404260225730077,"high":0.20452300485210667,"observations":36292,"sourceRank":2,"publishedAt":"2026-11-04","rank":1},{"name":"Claude Fable 5.1 (Max)","organization":"anthropic","score":0.15315588959346205,"low":0.13409416507074123,"high":0.17221761411618286,"observations":73015,"sourceRank":3,"publishedAt":"2026-11-04","rank":2},{"name":"GPT-6.1 Terra (Max)","organization":"openai","score":0.15182444034674886,"low":0.13799563584540997,"high":0.16565324484808774,"observations":118771,"sourceRank":4,"publishedAt":"2026-11-04","rank":2},{"name":"GPT-6 Astra (Max)","organization":"openai","score":0.14333657038536468,"low":0.12419200487202338,"high":0.162481135898706,"observations":47375,"sourceRank":5,"publishedAt":"2026-11-04","rank":3},{"name":"Gemini 4 Argon (High)","organization":"google","score":0.13518304059907676,"low":0.12558178856874155,"high":0.14478429262941198,"observations":262052,"sourceRank":6,"publishedAt":"2026-11-04","rank":3},{"name":"GPT-6.1 Sol (Max)","organization":"openai","score":0.13289150003053463,"low":0.12190133572197474,"high":0.1438816643390945,"observations":96131,"sourceRank":7,"publishedAt":"2026-11-04","rank":3},{"name":"Claude Sonnet 5.5 (Max)","organization":"anthropic","score":0.1239930914047766,"low":0.0948516922838564,"high":0.1531344905256968,"observations":17559,"sourceRank":8,"publishedAt":"2026-11-04","rank":3},{"name":"Muse Wonder Full (Max)","organization":"meta","score":0.10650764450790412,"low":0.09592412590345986,"high":0.11709116311234838,"observations":186836,"sourceRank":9,"publishedAt":"2026-11-04","rank":8},{"name":"GPT-6 Sol (Max)","organization":"openai","score":0.09983330889605269,"low":0.08272985243894071,"high":0.11693676535316466,"observations":67078,"sourceRank":10,"publishedAt":"2026-11-04","rank":8},{"name":"Claude Fable 5 (High)","organization":"anthropic","score":0.08792033856887327,"low":0.07415151382708438,"high":0.10168916331066216,"observations":167982,"sourceRank":11,"publishedAt":"2026-11-04","rank":8},{"name":"Hy4","organization":"tencent","score":0.06605197488285941,"low":0.04516607862535219,"high":0.08693787114036663,"observations":35444,"sourceRank":12,"publishedAt":"2026-11-04","rank":10},{"name":"Kimi K3.1 Pro (Max)","organization":"moonshot","score":0.0655115026941604,"low":0.049588003279871604,"high":0.08143500210844919,"observations":136888,"sourceRank":13,"publishedAt":"2026-11-04","rank":11},{"name":"GPT-5.6 Sol (xHigh)","organization":"openai","score":0.06341587906136442,"low":0.04953534100190494,"high":0.07729641712082391,"observations":166297,"sourceRank":14,"publishedAt":"2026-11-04","rank":11},{"name":"Claude Opus 5 (Max)","organization":"anthropic","score":0.053154912798258735,"low":0.03724587792950265,"high":0.06906394766701482,"observations":104038,"sourceRank":15,"publishedAt":"2026-11-04","rank":12},{"name":"GPT-6 Luna (Max)","organization":"openai","score":0.04194289523604683,"low":0.03621139271312329,"high":0.047674397758970366,"observations":1200732,"sourceRank":16,"publishedAt":"2026-11-04","rank":14},{"name":"Claude Sonnet 5 (High)","organization":"anthropic","score":0.040254461981612255,"low":0.023140722864437763,"high":0.05736820109878675,"observations":107206,"sourceRank":17,"publishedAt":"2026-11-04","rank":12},{"name":"DeepSeek-V4.1-Flash-20261017 (Max)","organization":"deepseek","score":0.034243770914607174,"low":0.026822466268765124,"high":0.041665075560449225,"observations":463497,"sourceRank":18,"publishedAt":"2026-11-04","rank":15},{"name":"Muse Spark 1.3 (Max)","organization":"meta","score":0.03398863118572004,"low":0.02293778689711826,"high":0.04503947547432183,"observations":473413,"sourceRank":19,"publishedAt":"2026-11-04","rank":15},{"name":"Claude Haiku 5.5 (High)","organization":"anthropic","score":0.03314158377449633,"low":0.020522497957101406,"high":0.045760669591891255,"observations":271177,"sourceRank":20,"publishedAt":"2026-11-04","rank":14},{"name":"Muse Wonder (Max)","organization":"meta","score":0.03122025319998345,"low":0.01305060804666176,"high":0.04938989835330514,"observations":109293,"sourceRank":21,"publishedAt":"2026-11-04","rank":14},{"name":"MiMo-V2.6-Flash","organization":"xiaomi","score":0.021373458300103676,"low":0.007530603968658331,"high":0.03521631263154902,"observations":125021,"sourceRank":22,"publishedAt":"2026-11-04","rank":17},{"name":"MiMo-V2.6-Pro","organization":"xiaomi","score":0.010053963136012478,"low":-0.00253669354434939,"high":0.022644619816374348,"observations":209174,"sourceRank":23,"publishedAt":"2026-11-04","rank":20},{"name":"GPT-5.6 Terra (xHigh)","organization":"openai","score":0.008166114110854328,"low":-0.002694908089807657,"high":0.019027136311516313,"observations":223809,"sourceRank":24,"publishedAt":"2026-11-04","rank":21},{"name":"GPT-6 Luna (Medium)","organization":"openai","score":0.0037866169800387928,"low":-0.0035370322043155227,"high":0.011110266164393107,"observations":469939,"sourceRank":25,"publishedAt":"2026-11-04","rank":22},{"name":"Kimi K3 (Max)","organization":"moonshot","score":0.0013451613261555861,"low":-0.007038369916357775,"high":0.009728692568668948,"observations":551540,"sourceRank":26,"publishedAt":"2026-11-04","rank":22},{"name":"Hy4 preview","organization":"tencent","score":0.0006774770208526348,"low":-0.009153275613397578,"high":0.010508229655102848,"observations":362751,"sourceRank":27,"publishedAt":"2026-11-04","rank":22},{"name":"Grok 4.6 (xHigh)","organization":"xai","score":0.00012900471974397476,"low":-0.016779165207082523,"high":0.017037174646570473,"observations":120399,"sourceRank":28,"publishedAt":"2026-11-04","rank":21},{"name":"Gemini 3.8 Flash (High)","organization":"google","score":-0.0024016685951619987,"low":-0.01410457637937906,"high":0.009301239189055063,"observations":267511,"sourceRank":29,"publishedAt":"2026-11-04","rank":22},{"name":"Qwen3.8-Max","organization":"alibaba","score":-0.0024995895183771433,"low":-0.012048212782272206,"high":0.00704903374551792,"observations":375140,"sourceRank":30,"publishedAt":"2026-11-04","rank":23},{"name":"GLM-5.2 (Max)","organization":"zai","score":-0.006041021029850269,"low":-0.017910707517049495,"high":0.0058286654573489576,"observations":287454,"sourceRank":31,"publishedAt":"2026-11-04","rank":23},{"name":"GLM-5.3 (Max)","organization":"zai","score":-0.008613496417297584,"low":-0.019922802078860234,"high":0.002695809244265067,"observations":489052,"sourceRank":32,"publishedAt":"2026-11-04","rank":23},{"name":"Step 5 Preview","organization":"stepfun","score":-0.00927405492081912,"low":-0.02818600689658229,"high":0.009637897054944052,"observations":67903,"sourceRank":33,"publishedAt":"2026-11-04","rank":22},{"name":"DeepSeek-V4-Pro (High, 20260813)","organization":"deepseek","score":-0.010766854088898326,"low":-0.03477816445608749,"high":0.013244456278290836,"observations":28886,"sourceRank":34,"publishedAt":"2026-11-04","rank":21},{"name":"Grok 4.7 (xHigh)","organization":"xai","score":-0.017537880654575067,"low":-0.03814118912347347,"high":0.003065427814323344,"observations":96104,"sourceRank":35,"publishedAt":"2026-11-04","rank":23},{"name":"Hy3","organization":"tencent","score":-0.027983181689711145,"low":-0.047967519988057816,"high":-0.007998843391364474,"observations":52740,"sourceRank":36,"publishedAt":"2026-11-04","rank":27},{"name":"GPT-6 Luna (Low)","organization":"openai","score":-0.028593965320671932,"low":-0.03849840627515348,"high":-0.01868952436619038,"observations":306594,"sourceRank":37,"publishedAt":"2026-11-04","rank":32},{"name":"Qwen3.8-Flash-Next","organization":"alibaba","score":-0.02959988043520878,"low":-0.04036785765943732,"high":-0.018831903210980244,"observations":372894,"sourceRank":38,"publishedAt":"2026-11-04","rank":32},{"name":"MiniMax-M3.1-Flash","organization":"minimax","score":-0.04099097834479182,"low":-0.06365832936715629,"high":-0.01832362732242734,"observations":43964,"sourceRank":39,"publishedAt":"2026-11-04","rank":32},{"name":"GLM-5.3-Flash","organization":"zai","score":-0.04234409354913371,"low":-0.053246365559523404,"high":-0.031441821538744016,"observations":682288,"sourceRank":40,"publishedAt":"2026-11-04","rank":34},{"name":"Qwen3.8-27B","organization":"alibaba","score":-0.050644788355669546,"low":-0.06449860255576381,"high":-0.03679097415557528,"observations":262266,"sourceRank":41,"publishedAt":"2026-11-04","rank":35},{"name":"Mistral Large 4","organization":"mistral","score":-0.05343913123733157,"low":-0.08403579370980542,"high":-0.022842468764857727,"observations":20703,"sourceRank":42,"publishedAt":"2026-11-04","rank":33},{"name":"Gemini 3.7 Flash (High)","organization":"google","score":-0.056880498746387564,"low":-0.07090408256512314,"high":-0.04285691492765199,"observations":297633,"sourceRank":43,"publishedAt":"2026-11-04","rank":38},{"name":"Gemini 3.1 Pro Preview","organization":"google","score":-0.07312927776916689,"low":-0.10026718474299699,"high":-0.045991370795336786,"observations":102576,"sourceRank":44,"publishedAt":"2026-11-04","rank":38},{"name":"Beam 1.0 501B A23B","organization":"reflection","score":-0.08216934892187466,"low":-0.11802195537677446,"high":-0.046316742466974876,"observations":44666,"sourceRank":45,"publishedAt":"2026-11-04","rank":38},{"name":"MiniMax-M3","organization":"minimax","score":-0.10642137866462387,"low":-0.12647540852549288,"high":-0.08636734880375485,"observations":86817,"sourceRank":46,"publishedAt":"2026-11-04","rank":44},{"name":"Spark-X2.6","organization":"synlan","score":-0.10934399073167167,"low":-0.13132921372621503,"high":-0.08735876773712829,"observations":44682,"sourceRank":47,"publishedAt":"2026-11-04","rank":44},{"name":"Qwen3.7-Max","organization":"alibaba","score":-0.11061481431756148,"low":-0.1325666030607767,"high":-0.08866302557434627,"observations":102402,"sourceRank":48,"publishedAt":"2026-11-04","rank":44},{"name":"Qwen3.7-Plus","organization":"alibaba","score":-0.11289328698622747,"low":-0.13959347290873653,"high":-0.08619310106371843,"observations":47955,"sourceRank":49,"publishedAt":"2026-11-04","rank":44},{"name":"Muse Spark 1.2 (xHigh)","organization":"meta","score":-0.11465899394692003,"low":-0.12930517632170407,"high":-0.10001281157213597,"observations":299841,"sourceRank":50,"publishedAt":"2026-11-04","rank":44},{"name":"Muse Spark 1.1","organization":"meta","score":-0.14137607272093244,"low":-0.1544337289096254,"high":-0.1283184165322395,"observations":473139,"sourceRank":51,"publishedAt":"2026-11-04","rank":47},{"name":"MiMo-V2.5-Pro","organization":"xiaomi","score":-0.14384625341781843,"low":-0.16430179062962394,"high":-0.12339071620601293,"observations":101141,"sourceRank":52,"publishedAt":"2026-11-04","rank":46},{"name":"Mistral Medium 3.5","organization":"mistral","score":-0.16234727681348016,"low":-0.19220966300811868,"high":-0.13248489061884164,"observations":38186,"sourceRank":53,"publishedAt":"2026-11-04","rank":49},{"name":"Solar Pro 4","organization":"upstage","score":-0.17011737317368902,"low":-0.20523768781441792,"high":-0.13499705853296012,"observations":33962,"sourceRank":54,"publishedAt":"2026-11-04","rank":50},{"name":"NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)","organization":"nvidia","score":-0.17135224674666738,"low":-0.18548244942166991,"high":-0.15722204407166485,"observations":117690,"sourceRank":55,"publishedAt":"2026-11-04","rank":52},{"name":"Nemotron 3 Ultra","organization":"nvidia","score":-0.18311786722374612,"low":-0.24384115195434156,"high":-0.12239458249315069,"observations":7695,"sourceRank":56,"publishedAt":"2026-11-04","rank":46},{"name":"Inkling","organization":"thinky","score":-0.20507492763948554,"low":-0.23876498429303852,"high":-0.17138487098593255,"observations":50902,"sourceRank":57,"publishedAt":"2026-11-04","rank":53},{"name":"Inkling-Small","organization":"thinky","score":-0.2575049328852109,"low":-0.29394564151471886,"high":-0.221064224255703,"observations":38000,"sourceRank":58,"publishedAt":"2026-11-04","rank":56}],"source":{"id":"arena","publisher":"Arena (lmarena-ai)","href":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset","licence":"CC-BY-4.0","licenceUrl":"https://creativecommons.org/licenses/by/4.0/","revision":"8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53","fetchedAt":"2026-11-05T06:35:05.520Z","evidence":[{"file":"a7a159f557a1e594a5d41218fe8d53c63c63daa39b655e51d6fc5b482a17d059.json","sha256":"a7a159f557a1e594a5d41218fe8d53c63c63daa39b655e51d6fc5b482a17d059","url":"https://huggingface.co/api/datasets/lmarena-ai/leaderboard-dataset","revision":null,"role":"metadata"},{"file":"88a46d472f35274fb6b7a852c3d16c8ee8007d51b37f290161537bec26895ee4.txt","sha256":"88a46d472f35274fb6b7a852c3d16c8ee8007d51b37f290161537bec26895ee4","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/raw/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/README.md","revision":"8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53","role":"licence"},{"file":"8eaf40bc3f60879f59b202aa8ac56a6e58716aa5110f48a1798ce43c63febf38.parquet","sha256":"8eaf40bc3f60879f59b202aa8ac56a6e58716aa5110f48a1798ce43c63febf38","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/agent_steerability/latest-00000-of-00001.parquet","revision":"8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53","role":"rows"}]},"status":"fresh"}