{"config":"agent_bash_recovery_steps","id":"arena-agent-bash-recovery-steps","title":"Agents · bash recovery (historical)","category":"Coding and agents","metric":"IPS score","direction":"higher","precision":6,"publishedAt":"2026-10-08","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.","Historical signal: Arena replaced bash recovery with bash reliability on 2026-10-12."],"omissions":[],"rows":[{"name":"Claude Sonnet 5.5 (Max)","organization":"anthropic","score":0.15035238150522567,"low":0.13332183348120594,"high":0.1673829295292454,"observations":9166,"sourceRank":1,"publishedAt":"2026-10-08","rank":1},{"name":"Claude Opus 5 (Max)","organization":"anthropic","score":0.13570115207168132,"low":0.1277785902477195,"high":0.14362371389564313,"observations":60324,"sourceRank":2,"publishedAt":"2026-10-08","rank":1},{"name":"Claude Opus 5.5 (High)","organization":"anthropic","score":0.1355335853279116,"low":0.12751138356278682,"high":0.14355578709303637,"observations":23322,"sourceRank":3,"publishedAt":"2026-10-08","rank":1},{"name":"Claude Opus 5 (High)","organization":"anthropic","score":0.12637498922659796,"low":0.11866892422755904,"high":0.13408105422563688,"observations":77004,"sourceRank":4,"publishedAt":"2026-10-08","rank":1},{"name":"Claude Fable 5.1 (Max)","organization":"anthropic","score":0.11603170489005432,"low":0.09287380763220318,"high":0.13918960214790546,"observations":36016,"sourceRank":5,"publishedAt":"2026-10-08","rank":1},{"name":"Claude Fable 5 (High)","organization":"anthropic","score":0.09597561240468977,"low":0.08375754359764019,"high":0.10819368121173935,"observations":49937,"sourceRank":6,"publishedAt":"2026-10-08","rank":5},{"name":"GPT 5.5 (xHigh)","organization":"openai","score":0.0954738256695622,"low":0.08396354616019815,"high":0.10698410517892624,"observations":54039,"sourceRank":7,"publishedAt":"2026-10-08","rank":5},{"name":"Deepseek V4.1 Flash (Max)","organization":"deepseek","score":0.0785815025696297,"low":0.07582897253381005,"high":0.08133403260544936,"observations":745424,"sourceRank":8,"publishedAt":"2026-10-08","rank":8},{"name":"Claude Sonnet 5 (High)","organization":"anthropic","score":0.073767542213248,"low":0.06222692478266662,"high":0.08530815964382937,"observations":125181,"sourceRank":9,"publishedAt":"2026-10-08","rank":6},{"name":"Claude Opus 4.8 (High)","organization":"anthropic","score":0.0722358959596161,"low":0.05321138752800306,"high":0.09126040439122915,"observations":40667,"sourceRank":10,"publishedAt":"2026-10-08","rank":6},{"name":"Grok 4.7 (xHigh)","organization":"xai","score":0.07203833245634916,"low":0.061656550677604165,"high":0.08242011423509416,"observations":41891,"sourceRank":11,"publishedAt":"2026-10-08","rank":8},{"name":"GPT 6 Sol (Max)","organization":"openai","score":0.07030190807524606,"low":0.05108683780259169,"high":0.08951697834790043,"observations":20341,"sourceRank":12,"publishedAt":"2026-10-08","rank":6},{"name":"Hy4 preview","organization":"tencent","score":0.06505790470733808,"low":0.06042575650204649,"high":0.06969005291262967,"observations":188996,"sourceRank":13,"publishedAt":"2026-10-08","rank":9},{"name":"Muse Spark 1.3 (Max)","organization":"meta","score":0.06077311225492221,"low":0.05481602128896373,"high":0.0667302032208807,"observations":146250,"sourceRank":14,"publishedAt":"2026-10-08","rank":9},{"name":"Muse Spark 1.2 (xHigh)","organization":"meta","score":0.05790054455603921,"low":0.048633440767251124,"high":0.0671676483448273,"observations":58077,"sourceRank":15,"publishedAt":"2026-10-08","rank":9},{"name":"Grok 4.6 (xHigh)","organization":"xai","score":0.05680396400430032,"low":0.046383945068269844,"high":0.0672239829403308,"observations":55194,"sourceRank":16,"publishedAt":"2026-10-08","rank":9},{"name":"GPT 6 Astra (Max)","organization":"openai","score":0.055274748105542004,"low":0.038102794216227906,"high":0.0724467019948561,"observations":24313,"sourceRank":17,"publishedAt":"2026-10-08","rank":9},{"name":"Inkling Small","organization":"thinky","score":0.05433972301988699,"low":0.03936850551916681,"high":0.06931094052060716,"observations":15423,"sourceRank":18,"publishedAt":"2026-10-08","rank":9},{"name":"GPT 6 Luna (Max)","organization":"openai","score":0.0520701982384284,"low":0.044084920435632204,"high":0.0600554760412246,"observations":305361,"sourceRank":19,"publishedAt":"2026-10-08","rank":12},{"name":"Grok 4.5","organization":"xai","score":0.05136066786226423,"low":0.03164929056965082,"high":0.07107204515487764,"observations":47481,"sourceRank":20,"publishedAt":"2026-10-08","rank":9},{"name":"GPT 6.1 Sol (Max)","organization":"openai","score":0.04528409239854958,"low":0.029377525507450743,"high":0.06119065928964842,"observations":15487,"sourceRank":21,"publishedAt":"2026-10-08","rank":11},{"name":"DeepSeek V4 Pro (High) (0813)","organization":"deepseek","score":0.04396717307829645,"low":0.031144418898012016,"high":0.05678992725858088,"observations":23366,"sourceRank":22,"publishedAt":"2026-10-08","rank":12},{"name":"Kimi K3 (Max)","organization":"moonshot","score":0.04379465536790916,"low":0.038320163574305684,"high":0.04926914716151264,"observations":341546,"sourceRank":23,"publishedAt":"2026-10-08","rank":15},{"name":"GPT 5.5","organization":"openai","score":0.041197741522889464,"low":0.02187508880241374,"high":0.06052039424336519,"observations":53734,"sourceRank":24,"publishedAt":"2026-10-08","rank":11},{"name":"Qwen3.8 Max","organization":"alibaba","score":0.027983391648861877,"low":0.02174293935191649,"high":0.03422384394580727,"observations":143441,"sourceRank":25,"publishedAt":"2026-10-08","rank":21},{"name":"MiMo V2.6 Pro","organization":"xiaomi","score":0.023425984954652757,"low":0.0030873574716277427,"high":0.04376461243767777,"observations":28437,"sourceRank":26,"publishedAt":"2026-10-08","rank":18},{"name":"mistral-large-4","organization":"mistral","score":0.017771044807505465,"low":-0.012688299671309126,"high":0.048230389286320056,"observations":6001,"sourceRank":27,"publishedAt":"2026-10-08","rank":16},{"name":"GPT 5.4 (High)","organization":"openai","score":0.012145414242588082,"low":-0.0017805430894989784,"high":0.026071371574675144,"observations":85004,"sourceRank":28,"publishedAt":"2026-10-08","rank":24},{"name":"GPT 5.6 Terra (xHigh)","organization":"openai","score":0.008842784639766732,"low":-0.00867734886040105,"high":0.026362918139934514,"observations":48987,"sourceRank":29,"publishedAt":"2026-10-08","rank":24},{"name":"GPT 5.6 Sol (xHigh)","organization":"openai","score":0.0072627731198375844,"low":-0.006523497845149563,"high":0.021049044084824733,"observations":95365,"sourceRank":30,"publishedAt":"2026-10-08","rank":26},{"name":"GLM 5.2 (Max)","organization":"zai","score":0.0067463111759968886,"low":-0.0013236356710550502,"high":0.014816258023048827,"observations":215429,"sourceRank":31,"publishedAt":"2026-10-08","rank":26},{"name":"Minimax M3","organization":"minimax","score":0.005799036305831118,"low":-0.006340998917306889,"high":0.017939071528969126,"observations":89546,"sourceRank":32,"publishedAt":"2026-10-08","rank":26},{"name":"GPT 5.6 Luna (xHigh)","organization":"openai","score":0.005144852705668113,"low":-0.005483332601142353,"high":0.015773038012478577,"observations":108092,"sourceRank":33,"publishedAt":"2026-10-08","rank":26},{"name":"Step 5 Preview","organization":"stepfun","score":0.002284190481464776,"low":-0.013767974931968487,"high":0.01833635589489804,"observations":45610,"sourceRank":34,"publishedAt":"2026-10-08","rank":26},{"name":"Inkling","organization":"thinky","score":0.0008143083339426926,"low":-0.0242902180732562,"high":0.025918834741141584,"observations":46731,"sourceRank":35,"publishedAt":"2026-10-08","rank":24},{"name":"GLM 5.3 (Max)","organization":"zai","score":-0.0005353086799178008,"low":-0.006956233312116849,"high":0.005885615952281247,"observations":362858,"sourceRank":36,"publishedAt":"2026-10-08","rank":26},{"name":"Gemini 3.8 Flash (High)","organization":"google","score":-0.006867809731099461,"low":-0.014029149382367297,"high":0.0002935299201683745,"observations":248198,"sourceRank":37,"publishedAt":"2026-10-08","rank":27},{"name":"MiMo V2.6 Flash","organization":"xiaomi","score":-0.010597459793076691,"low":-0.031132050512391738,"high":0.009937130926238356,"observations":63663,"sourceRank":38,"publishedAt":"2026-10-08","rank":26},{"name":"Qwen3.7 Max","organization":"alibaba","score":-0.01560714420881748,"low":-0.031816984763701275,"high":0.0006026963460663178,"observations":52845,"sourceRank":39,"publishedAt":"2026-10-08","rank":27},{"name":"Muse Spark 1.1","organization":"meta","score":-0.02258230890923607,"low":-0.033950634781575915,"high":-0.011213983036896226,"observations":162956,"sourceRank":40,"publishedAt":"2026-10-08","rank":34},{"name":"Qwen3.8 Flash Next","organization":"alibaba","score":-0.025728882638138995,"low":-0.03072538273351874,"high":-0.020732382542759248,"observations":547513,"sourceRank":41,"publishedAt":"2026-10-08","rank":37},{"name":"Gemini 3.7 Flash (High)","organization":"google","score":-0.05443636647576189,"low":-0.0657906442461899,"high":-0.04308208870533388,"observations":237195,"sourceRank":42,"publishedAt":"2026-10-08","rank":42},{"name":"GLM 5.3 Flash","organization":"zai","score":-0.05699069676314816,"low":-0.06327907846402586,"high":-0.05070231506227046,"observations":412056,"sourceRank":43,"publishedAt":"2026-10-08","rank":42},{"name":"Qwen 3.8 27B","organization":"alibaba","score":-0.059014950436783176,"low":-0.0655214256361273,"high":-0.052508475237439046,"observations":287532,"sourceRank":44,"publishedAt":"2026-10-08","rank":42},{"name":"Qwen3.7 Plus","organization":"alibaba","score":-0.0630996816869549,"low":-0.09319579492404513,"high":-0.03300356844986467,"observations":34121,"sourceRank":45,"publishedAt":"2026-10-08","rank":41},{"name":"Hy3","organization":"tencent","score":-0.07205807009961007,"low":-0.09933702253172928,"high":-0.04477911766749086,"observations":29130,"sourceRank":46,"publishedAt":"2026-10-08","rank":42},{"name":"Mimo V2.5 Pro","organization":"xiaomi","score":-0.07505049308821143,"low":-0.09063443137989208,"high":-0.059466554796530784,"observations":69554,"sourceRank":47,"publishedAt":"2026-10-08","rank":42},{"name":"Gemini 3.6 Flash (High)","organization":"google","score":-0.18225355243790564,"low":-0.20278892623279066,"high":-0.16171817864302063,"observations":77980,"sourceRank":48,"publishedAt":"2026-10-08","rank":48},{"name":"Mistral Medium 3.5","organization":"mistral","score":-0.20447688894985472,"low":-0.23825590640423971,"high":-0.17069787149546972,"observations":18687,"sourceRank":49,"publishedAt":"2026-10-08","rank":48},{"name":"Gemini 4 Argon (High)","organization":"google","score":-0.21405741163545722,"low":-0.2520289834401014,"high":-0.17608583983081305,"observations":39438,"sourceRank":50,"publishedAt":"2026-10-08","rank":48},{"name":"Gemini 3.1 Pro Preview","organization":"google","score":-0.23265218021387013,"low":-0.2513309624549065,"high":-0.2139733979728338,"observations":224173,"sourceRank":51,"publishedAt":"2026-10-08","rank":49},{"name":"Solar Pro 4","organization":"upstage","score":-0.25557417868208243,"low":-0.30070562107506127,"high":-0.2104427362891036,"observations":23430,"sourceRank":52,"publishedAt":"2026-10-08","rank":49}],"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":"07fa4c2c648f685e4f8bf93a54b34fff4cbdacfc12c2d34f77abd9cde3c0191d.parquet","sha256":"07fa4c2c648f685e4f8bf93a54b34fff4cbdacfc12c2d34f77abd9cde3c0191d","url":"https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset/resolve/8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53/agent_bash_recovery_steps/latest-00000-of-00001.parquet","revision":"8b6deb69ace01ee8dcda9e46eba7a5e8a808ec53","role":"rows"}]},"status":"fresh"}