[{"data":1,"prerenderedAt":331},["ShallowReactive",2],{"site-en":3,"posts-en":79,"alt-\u002Fen\u002Fai\u002F":326,"graphic-\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fdaisyrose-27b-benchmark-results.svg":329,"graphic-\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark.svg":330},{"brand":4,"nav":8,"home":18,"story":25,"footer":37,"share":39,"locale":49,"sections":55,"section":65,"editions":69},{"name":5,"domain":6,"tagline":7},"Infografx","infografx.org","What is changing in AI, finance and law — in numbers you can check.",[9,12,15],{"label":10,"to":11},"AI","ai\u002F",{"label":13,"to":14},"Finance","finance\u002F",{"label":16,"to":17},"Law","legal\u002F",{"title":19,"description":20,"leadKicker":21,"moreTitle":22,"readLabel":23,"questionLabel":24},"Infografx — AI, finance and law in verified numbers","Short data stories and original infographics on AI, finance and law. Every number links to its primary source.","Lead story","More stories","Read the story","The question",{"questionLabel":24,"sourceLabel":26,"unitLabel":27,"periodLabel":28,"limitationsLabel":29,"sourcesLabel":30,"primaryLabel":31,"contextLabel":32,"moreLabel":22,"readTime":33,"eventLabel":34,"publishedLabel":35,"whyLabel":36},"Source","Unit","Period","Read this before you use the numbers","Sources","Primary","Context","min read","Event date","Published","Why it matters.",{"note":38},"Infografx — independent data journalism on AI, finance and law.",{"copy":40,"copied":41,"download":42,"linkedin":43,"native":44,"more":45,"story":46,"og":47,"storyCta":48},"Copy link","Link copied","Download chart","LinkedIn","Share","More formats","Vertical 9:16 — TikTok, Stories","Link card 1.91:1","Full story and sources",{"code":50,"htmlLang":50,"ogLocale":51,"label":52,"name":53,"dateLocale":54},"en","en_GB","EN","English","en-GB",{"ai":56,"finance":59,"legal":62},{"label":10,"accent":57,"description":58},"ai","Data stories on AI models, compute and adoption.",{"label":13,"accent":60,"description":61},"finance","Data stories on markets, earnings, rates and prices.",{"label":16,"accent":63,"description":64},"law","Data stories on rulings, enforcement and regulation.",{"kicker":66,"storiesLabel":67,"empty":68},"Section","stories","No stories yet.",{"label":70,"items":71},"Edition",[72,73,76],{"code":50,"label":53},{"code":74,"label":75},"fr","Français",{"code":77,"label":78},"en-ae","UAE",[80,133,180,227,283],{"title":81,"description":82,"path":83,"niche":60,"slug":84,"key":84,"locale":50,"category":60,"publishedAt":85,"updatedAt":85,"eventDate":86,"lead":87,"question":88,"graphic":89,"graphicAlt":90,"images":91,"chart":95,"sources":117,"limitations":131,"readTimeMinutes":132},"Nvidia data center revenue rises to $107.6 billion","Nvidia reported $107.6 billion in quarterly data center revenue, with $47.5 billion from AI clouds, industrial customers and enterprises.","\u002Fen\u002Ffinance\u002Fnvidia-data-center-revenue-q3-fy27","nvidia-data-center-revenue-q3-fy27","2026-11-19","2026-11-17",true,"How did Nvidia's two data center customer groups grow over a year?","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fnvidia-data-center-revenue-q3-fy27.svg","Hyperscale revenue rose from $35.1 billion to $60.1 billion. AI clouds, industrial and enterprise revenue rose from $16.1 billion to $47.5 billion.",{"feed":92,"story":93,"og":94},"\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fnvidia-data-center-revenue-q3-fy27-feed.png","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fnvidia-data-center-revenue-q3-fy27-story.png","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fnvidia-data-center-revenue-q3-fy27-og.png",{"type":96,"title":97,"takeaway":98,"whyItMatters":99,"unit":100,"period":101,"series":102,"highlight":113,"source":115,"sourceUrl":116},"bar","Nvidia’s non-hyperscale sales nearly tripled","Hyperscale remained larger, while AI clouds, industrial and enterprise revenue grew more in dollar terms.","The breakdown shows where Nvidia's data center revenue growth came from.","USD billion","Q3 FY26 and Q3 FY27",[103,106,109,112],{"label":104,"value":105},"Hyperscale · Q3 FY26",35.1,{"label":107,"value":108},"Hyperscale · Q3 FY27",60.1,{"label":110,"value":111},"AI clouds, industrial and enterprise · Q3 FY26",16.1,{"label":113,"value":114},"AI clouds, industrial and enterprise · Q3 FY27",47.5,"Nvidia — Q3 FY27 CFO commentary (SEC Form 8-K)","https:\u002F\u002Fwww.sec.gov\u002FArchives\u002Fedgar\u002Fdata\u002F1045810\u002F000104581026000088\u002Fq3fy27cfocommentary.htm",[118,122,124,127],{"name":119,"url":120,"date":86,"type":121},"Nvidia — Q3 FY27 quarterly presentation","https:\u002F\u002Finvestor.nvidia.com\u002Ffiles\u002Fdoc_financials\u002F2027\u002FQ327\u002FNVDA-F3Q27-Quarterly-Presentation.pdf","primary",{"name":123,"url":116,"date":86,"type":121},"Nvidia — CFO commentary, Q3 FY27 (SEC Form 8-K, exhibit 99.2)",{"name":125,"url":126,"date":86,"type":121},"Nvidia — Third-quarter fiscal 2027 results","https:\u002F\u002Finvestor.nvidia.com\u002Fnews\u002Fpress-release-details\u002F2026\u002FNVIDIA-Announces-Financial-Results-for-Third-Quarter-Fiscal-2027\u002F",{"name":128,"url":129,"date":86,"type":130},"Reddit — Nvidia Q3 fiscal 2027 results discussion","https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fstocks\u002Fcomments\u002F1xywpnl\u002Fnvidia_announces_financial_results_for_third\u002F","context","Nvidia reports these customer-group figures in billions, to one decimal place, and combines AI clouds, industrial customers and enterprises into one group.",1,{"title":134,"description":135,"path":136,"niche":57,"slug":137,"key":137,"locale":50,"category":57,"publishedAt":85,"updatedAt":85,"eventDate":138,"lead":139,"question":140,"graphic":141,"graphicAlt":142,"images":143,"chart":147,"sources":174,"limitations":179,"readTimeMinutes":132},"DaisyRose-27B scores 12% on Terminal-Bench 4.1","RadixArk released DaisyRose-27B and reported a 12.0% score on Terminal-Bench 4.1, up from 5.6% for the base model.","\u002Fen\u002Fai\u002Fdaisyrose-27b-benchmark-results","daisyrose-27b-benchmark-results","2026-11-18",false,"How did DaisyRose-27B score against its base model on three tests?","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fdaisyrose-27b-benchmark-results.svg","Qwen and DaisyRose scored 5.6% and 12.0% on Terminal-Bench 4.1, 73.0% and 75.5% on Terminal-Bench 2.1, and 42.2% and 46.2% on DeepSWE.",{"feed":144,"story":145,"og":146},"\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fdaisyrose-27b-benchmark-results-feed.png","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fdaisyrose-27b-benchmark-results-story.png","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fdaisyrose-27b-benchmark-results-og.png",{"type":96,"title":148,"takeaway":149,"whyItMatters":150,"unit":151,"period":152,"series":153,"highlight":158,"source":172,"sourceUrl":173},"DaisyRose gains most on Terminal-Bench 4.1","DaisyRose beat the base model on each test shown, with its widest gap on Terminal-Bench 4.1.","The paired scores show where the model's later training made the largest difference.","percent","Results released 18 November 2026",[154,157,160,163,166,169],{"label":155,"value":156},"Terminal-Bench 4.1 · Qwen",5.6,{"label":158,"value":159},"Terminal-Bench 4.1 · DaisyRose",12,{"label":161,"value":162},"Terminal-Bench 2.1 · Qwen",73,{"label":164,"value":165},"Terminal-Bench 2.1 · DaisyRose",75.5,{"label":167,"value":168},"DeepSWE v1.1 · Qwen",42.2,{"label":170,"value":171},"DeepSWE v1.1 · DaisyRose",46.2,"RadixArk — DaisyRose-27B release and benchmark table","https:\u002F\u002Fhuggingface.co\u002Fblog\u002FRadixArk\u002Fdaisyrose-27b-pushing-the-frontier-of-post-trainin",[175,176],{"name":172,"url":173,"date":138,"type":121},{"name":177,"url":178,"date":138,"type":130},"Hacker News — DaisyRose-27B discussion","https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=50551046","RadixArk ran these tests. Compare the models within each test because the tests use different tasks.",{"title":181,"description":182,"path":183,"niche":60,"slug":184,"key":184,"locale":50,"category":60,"publishedAt":85,"updatedAt":85,"eventDate":138,"lead":139,"question":185,"graphic":186,"graphicAlt":187,"images":188,"chart":192,"sources":221,"limitations":226,"readTimeMinutes":132},"UK inflation reaches 3.6% in October","UK consumer prices rose 3.6% in the year to October, up from 3.5% in September.","\u002Fen\u002Ffinance\u002Fuk-inflation-october-2026","uk-inflation-october-2026","How did the UK's annual consumer price inflation rate change since April?","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fuk-inflation-october-2026.svg","UK annual CPI inflation was 2.8% in April and May, fell to 2.6% in June, then rose each month to 3.6% in October.",{"feed":189,"story":190,"og":191},"\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fuk-inflation-october-2026-feed.png","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fuk-inflation-october-2026-story.png","\u002Fcontent\u002Fen\u002Fimg\u002Ffinance\u002Fuk-inflation-october-2026-og.png",{"type":193,"title":194,"takeaway":195,"whyItMatters":196,"unit":151,"period":197,"series":198,"highlight":217,"source":219,"sourceUrl":220},"column","UK inflation rose each month after June","Annual CPI inflation rose from 2.6% in June to 3.6% in October.","The annual rate shows how quickly prices changed compared with the same month a year earlier.","April to October 2026",[199,202,204,207,210,213,216],{"label":200,"value":201},"Apr 2026",2.8,{"label":203,"value":201},"May 2026",{"label":205,"value":206},"Jun 2026",2.6,{"label":208,"value":209},"Jul 2026",2.9,{"label":211,"value":212},"Aug 2026",3.1,{"label":214,"value":215},"Sep 2026",3.5,{"label":217,"value":218},"Oct 2026",3.6,"Office for National Statistics — Consumer price inflation, UK: October 2026","https:\u002F\u002Fwww.ons.gov.uk\u002Feconomy\u002Finflationandpriceindices\u002Fbulletins\u002Fconsumerpriceinflation\u002Foctober2026",[222,223],{"name":219,"url":220,"date":138,"type":121},{"name":224,"url":225,"date":138,"type":130},"Reddit — October UK inflation discussion","https:\u002F\u002Fwww.reddit.com\u002Fr\u002Fukpolitics\u002Fcomments\u002F1xz8u4n\u002Fuk_inflation_rate_rises_again_hitting_36_in\u002F","The CPI uses a national basket of goods and services. A household's costs can change at a different rate.",{"title":228,"description":229,"path":230,"niche":231,"slug":232,"key":232,"locale":50,"category":63,"publishedAt":85,"updatedAt":85,"eventDate":138,"lead":139,"question":233,"graphic":234,"graphicAlt":235,"images":236,"chart":240,"sources":258,"limitations":282,"readTimeMinutes":132},"UK regulator opens three investigations into hidden fees","The UK competition regulator announced three investigations into whether travel and gym firms showed mandatory fees upfront.","\u002Fen\u002Flegal\u002Fcma-drip-pricing-travel-and-gyms","legal","cma-drip-pricing-travel-and-gyms","How much did the CMA fine companies in three earlier drip pricing settlements?","\u002Fcontent\u002Fen\u002Fimg\u002Flegal\u002Fcma-drip-pricing-travel-and-gyms.svg","Earlier CMA drip pricing fines were £4,200,000 for the AA, £889,200 for StubHub UK and £45,000 for Gold’s Gym.",{"feed":237,"story":238,"og":239},"\u002Fcontent\u002Fen\u002Fimg\u002Flegal\u002Fcma-drip-pricing-travel-and-gyms-feed.png","\u002Fcontent\u002Fen\u002Fimg\u002Flegal\u002Fcma-drip-pricing-travel-and-gyms-story.png","\u002Fcontent\u002Fen\u002Fimg\u002Flegal\u002Fcma-drip-pricing-travel-and-gyms-og.png",{"type":96,"title":241,"takeaway":242,"whyItMatters":243,"unit":244,"period":245,"series":246,"highlight":248,"source":256,"sourceUrl":257},"Earlier drip pricing fines ranged from £45,000 to £4.2 million","The AA’s £4.2 million penalty was the largest of the three earlier settlements shown.","Earlier settlements show how the CMA has used its direct enforcement powers.","GBP","Decisions from April to September 2026",[247,250,253],{"label":248,"value":249},"AA and BSM driving schools",4200000,{"label":251,"value":252},"StubHub UK",889200,{"label":254,"value":255},"Gold’s Gym",45000,"Competition and Markets Authority, three enforcement case records","https:\u002F\u002Fwww.gov.uk\u002Fcma-cases\u002Fautomobile-association-developments-limited-consumer-protection-enforcement-case",[259,262,265,268,272,276,279],{"name":260,"url":261,"date":138,"type":121},"Competition and Markets Authority — Travel and leisure businesses investigated over pricing practices","https:\u002F\u002Fwww.gov.uk\u002Fgovernment\u002Fnews\u002Ftravel-and-leisure-businesses-investigated-over-pricing-practices",{"name":263,"url":257,"date":264,"type":121},"Competition and Markets Authority — Automobile Association Developments Limited case record","2026-04-15",{"name":266,"url":267,"date":264,"type":121},"Competition and Markets Authority — Automobile Association Developments Limited final infringement notice","https:\u002F\u002Fassets.publishing.service.gov.uk\u002Fmedia\u002F69fb242c8a56cc5f2174fa99\u002FNon-confidential_Final_Infringement_Notice_6.5.26.pdf",{"name":269,"url":270,"date":271,"type":121},"Competition and Markets Authority — StubHub UK case record","https:\u002F\u002Fwww.gov.uk\u002Fcma-cases\u002Fstubhub-consumer-protection-enforcement-case","2026-06-23",{"name":273,"url":274,"date":275,"type":121},"Competition and Markets Authority — Gold’s Gym case record","https:\u002F\u002Fwww.gov.uk\u002Fcma-cases\u002Fgolds-gym-uk-limited-consumer-protection-enforcement-case","2026-09-29",{"name":277,"url":278,"date":275,"type":121},"Competition and Markets Authority — Gold’s Gym final infringement notice","https:\u002F\u002Fassets.publishing.service.gov.uk\u002Fmedia\u002F6ace3abbb8542d3693fd4403\u002F___Final_Infringement_Notice__.pdf",{"name":280,"url":281,"date":138,"type":130},"Slaughter and May — Commentary on the new drip pricing investigations","https:\u002F\u002Fthelens.slaughterandmay.com\u002Fpost\u002F102oa62\u002Fcma-enforcement-drive-continues-with-three-more-drip-pricing-investigations","The chart shows penalties after discounts in three past settlements. The CMA has made no finding of infringement in the new cases.",{"title":284,"description":285,"path":286,"niche":57,"slug":287,"key":287,"locale":50,"category":57,"publishedAt":85,"updatedAt":85,"eventDate":86,"lead":139,"question":288,"graphic":289,"graphicAlt":290,"images":291,"chart":295,"sources":311,"limitations":325,"readTimeMinutes":132},"GPT-6.2 Astra scores 62.5% on repeat financial-analysis tests","OpenAI released GPT-6.2 Astra and Sol, and Astra answered 62.5% of the benchmark questions correctly on every attempt at its two highest tested settings.","\u002Fen\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark","gpt-6-2-astra-financial-analysis-benchmark","How often did GPT-6.2 Astra answer correctly on all five attempts?","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark.svg","On AA-AnalystAgent, GPT-6.2 Astra scored 62.5% at Max and Xhigh and 61.3% at High.",{"feed":292,"story":293,"og":294},"\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark-feed.png","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark-story.png","\u002Fcontent\u002Fen\u002Fimg\u002Fai\u002Fgpt-6-2-astra-financial-analysis-benchmark-og.png",{"type":96,"title":296,"takeaway":297,"whyItMatters":298,"unit":151,"period":299,"series":300,"source":309,"sourceUrl":310},"62.5% right all five times","Max and Xhigh scored 62.5%, and High scored 61.3%.","The test measures how often answers stay correct across repeated attempts.","Leaderboard as of 19 November 2026",[301,304,306],{"label":302,"value":303},"Max",62.5,{"label":305,"value":303},"Xhigh",{"label":307,"value":308},"High",61.3,"Artificial Analysis — AA-AnalystAgent Benchmark","https:\u002F\u002Fartificialanalysis.ai\u002Fevaluations\u002Faa-analyst-agent",[312,313,316,319,322],{"name":309,"url":310,"date":85,"type":121},{"name":314,"url":315,"date":86,"type":130},"Artificial Analysis — GPT-6.2 Astra model page (release date)","https:\u002F\u002Fartificialanalysis.ai\u002Fmodels\u002Fgpt-6-2-astra",{"name":317,"url":318,"date":86,"type":121},"OpenAI — Introducing GPT-6.2 Astra and Sol","https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-gpt-6-2-astra-sol\u002F",{"name":320,"url":321,"date":86,"type":130},"Reddit — GPT-6.2 Astra and Sol benchmark discussion","https:\u002F\u002Fwww.reddit.com\u002Fr\u002FOpenAI\u002Fcomments\u002F1xyuzly\u002Fgpt62_astra_sol_benchmarks_openal_nailed_the\u002F",{"name":323,"url":324,"date":86,"type":130},"Hacker News — Artificial Analysis benchmarks for GPT-6.2 Astra","https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=50535783","The benchmark uses a fixed set of source files and questions. The scores apply to those tasks and test settings.",{"en":327,"fr":328},"\u002Fen\u002Fai\u002F","\u002Ffr\u002Fia\u002F","\u003Csvg xmlns=\"http:\u002F\u002Fwww.w3.org\u002F2000\u002Fsvg\" viewBox=\"0 0 1080 1350\" role=\"img\" aria-labelledby=\"daisyrose-27b-benchmark-results-t daisyrose-27b-benchmark-results-d\">\n\u003Ctitle id=\"daisyrose-27b-benchmark-results-daisyrose-27b-benchmark-results-t\">DaisyRose gains most on Terminal-Bench 4.1\u003C\u002Ftitle>\u003Cdesc id=\"daisyrose-27b-benchmark-results-daisyrose-27b-benchmark-results-d\">Qwen and DaisyRose scored 5.6% and 12.0% on Terminal-Bench 4.1, 73.0% and 75.5% on Terminal-Bench 2.1, and 42.2% and 46.2% on DeepSWE.\u003C\u002Fdesc>\n\u003Crect width=\"1080\" height=\"1350\" fill=\"#f4eee3\"\u002F>\n\u003Cg transform=\"translate(60 62) scale(1.375)\">\u003Crect x=\"2\" y=\"17\" width=\"6\" height=\"13\" rx=\"1.2\" fill=\"#0e6f97\"\u002F>\u003Crect x=\"13\" y=\"9\" width=\"6\" height=\"21\" rx=\"1.2\" fill=\"#3e7a1d\"\u002F>\u003Crect x=\"24\" y=\"2\" width=\"6\" height=\"28\" rx=\"1.2\" fill=\"#9b7424\"\u002F>\u003C\u002Fg>\n\u003Ctext x=\"118\" y=\"99\" font-family=\"'Fraunces Variable', Georgia, serif\" font-weight=\"600\" font-size=\"36\" fill=\"#1b1915\">Infografx\u003C\u002Ftext>\n\u003Ctext x=\"1020\" y=\"96\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"24\" letter-spacing=\"3\" fill=\"#0e6f97\" text-anchor=\"end\">AI\u003C\u002Ftext>\n\u003Crect x=\"60\" y=\"132\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\n\u003Ctext font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"26\" letter-spacing=\"2\" fill=\"#6e675d\">\u003Ctspan x=\"60\" y=\"192\">HOW DID DAISYROSE-27B SCORE AGAINST ITS BASE MODEL\u003C\u002Ftspan>\u003Ctspan x=\"60\" y=\"230\">ON THREE TESTS?\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext font-family=\"'Fraunces Variable', Georgia, serif\" font-weight=\"600\" font-size=\"68\" letter-spacing=\"-0.5\" fill=\"#1b1915\">\u003Ctspan x=\"60\" y=\"322\">DaisyRose gains most on\u003C\u002Ftspan>\u003Ctspan x=\"60\" y=\"398\">Terminal-Bench 4.1\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Crect x=\"60\" y=\"1070\" width=\"4\" height=\"84\" fill=\"#9b7424\"\u002F>\n\u003Ctext font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"30\" fill=\"#45403a\">\u003Ctspan x=\"88\" y=\"1100\">The paired scores show where the model’s later\u003C\u002Ftspan>\u003Ctspan x=\"88\" y=\"1142\">training made the largest difference.\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"22\" fill=\"#6e675d\">\u003Ctspan x=\"60\" y=\"1204\">Source: RadixArk — DaisyRose-27B release and benchmark table · Results\u003C\u002Ftspan>\u003Ctspan x=\"60\" y=\"1236\">released 18 November 2026 · Company-reported · Unit: percent\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext x=\"60\" y=\"472\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"22\" fill=\"#6e675d\">QWEN = BASE MODEL QWEN3.8-27B · SCORE, %\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"544\" font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"30\" font-weight=\"600\" fill=\"#1b1915\">Terminal-Bench 4.1\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"586.3333333333334\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#6e675d\">Qwen\u003C\u002Ftext>\u003Crect x=\"260\" y=\"560\" width=\"289.3\" height=\"34.66666666666667\" rx=\"3\" fill=\"#cfc4b1\" data-label=\"Terminal-Bench 4.1 · Qwen\" data-value=\"5.6\"\u002F>\u003Ctext x=\"565.3\" y=\"588.1333333333333\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">5.6%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"631\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#1b1915\">DaisyRose\u003C\u002Ftext>\u003Crect x=\"260\" y=\"604.6666666666666\" width=\"620.0\" height=\"34.66666666666667\" rx=\"3\" fill=\"#0e6f97\" data-label=\"Terminal-Bench 4.1 · DaisyRose\" data-value=\"12\"\u002F>\u003Ctext x=\"896.0\" y=\"632.8\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#0e6f97\">12.0%\u003C\u002Ftext>\u003Crect x=\"60\" y=\"652.3333333333334\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\u003Ctext x=\"60\" y=\"703.3333333333334\" font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"30\" font-weight=\"600\" fill=\"#1b1915\">Terminal-Bench 2.1\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"745.6666666666667\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#6e675d\">Qwen\u003C\u002Ftext>\u003Crect x=\"260\" y=\"719.3333333333334\" width=\"599.5\" height=\"34.66666666666667\" rx=\"3\" fill=\"#cfc4b1\" data-label=\"Terminal-Bench 2.1 · Qwen\" data-value=\"73\"\u002F>\u003Ctext x=\"875.5\" y=\"747.4666666666667\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">73.0%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"790.3333333333334\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#6e675d\">DaisyRose\u003C\u002Ftext>\u003Crect x=\"260\" y=\"764\" width=\"620.0\" height=\"34.66666666666667\" rx=\"3\" fill=\"#0e6f97\" fill-opacity=\"0.35\" data-label=\"Terminal-Bench 2.1 · DaisyRose\" data-value=\"75.5\"\u002F>\u003Ctext x=\"896.0\" y=\"792.1333333333333\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">75.5%\u003C\u002Ftext>\u003Crect x=\"60\" y=\"811.6666666666667\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\u003Ctext x=\"60\" y=\"862.6666666666667\" font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"30\" font-weight=\"600\" fill=\"#1b1915\">DeepSWE v1.1\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"905.0000000000001\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#6e675d\">Qwen\u003C\u002Ftext>\u003Crect x=\"260\" y=\"878.6666666666667\" width=\"566.3\" height=\"34.66666666666667\" rx=\"3\" fill=\"#cfc4b1\" data-label=\"DeepSWE v1.1 · Qwen\" data-value=\"42.2\"\u002F>\u003Ctext x=\"842.3\" y=\"906.8000000000001\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">42.2%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"949.6666666666667\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"28\" fill=\"#6e675d\">DaisyRose\u003C\u002Ftext>\u003Crect x=\"260\" y=\"923.3333333333334\" width=\"620.0\" height=\"34.66666666666667\" rx=\"3\" fill=\"#0e6f97\" fill-opacity=\"0.35\" data-label=\"DeepSWE v1.1 · DaisyRose\" data-value=\"46.2\"\u002F>\u003Ctext x=\"896.0\" y=\"951.4666666666667\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">46.2%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"1028\" font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"24\" fill=\"#6e675d\">Each test has its own scale. Compare scores within each test.\u003C\u002Ftext>\n\u003Crect x=\"0\" y=\"1266\" width=\"1080\" height=\"84\" fill=\"#f4eee3\"\u002F>\n\u003Crect x=\"60\" y=\"1266\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\n\u003Cg transform=\"translate(60 1282) scale(0.9375)\">\u003Crect x=\"2\" y=\"17\" width=\"6\" height=\"13\" rx=\"1.2\" fill=\"#0e6f97\"\u002F>\u003Crect x=\"13\" y=\"9\" width=\"6\" height=\"21\" rx=\"1.2\" fill=\"#3e7a1d\"\u002F>\u003Crect x=\"24\" y=\"2\" width=\"6\" height=\"28\" rx=\"1.2\" fill=\"#9b7424\"\u002F>\u003C\u002Fg>\n\u003Ctext x=\"104\" y=\"1307\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"26\" fill=\"#1b1915\" fill-opacity=\"0.7\">infografx.org\u003C\u002Ftext>\n\u003C\u002Fsvg>\n","\u003Csvg xmlns=\"http:\u002F\u002Fwww.w3.org\u002F2000\u002Fsvg\" viewBox=\"0 0 1080 1350\" role=\"img\" aria-labelledby=\"gpt-6-2-astra-financial-analysis-benchmark-t gpt-6-2-astra-financial-analysis-benchmark-d\">\n\u003Ctitle id=\"gpt-6-2-astra-financial-analysis-benchmark-gpt-6-2-astra-financial-analysis-benchmark-t\">62.5% right all five times\u003C\u002Ftitle>\u003Cdesc id=\"gpt-6-2-astra-financial-analysis-benchmark-gpt-6-2-astra-financial-analysis-benchmark-d\">On AA-AnalystAgent, GPT-6.2 Astra scored 62.5% at Max and Xhigh and 61.3% at High.\u003C\u002Fdesc>\n\u003Crect width=\"1080\" height=\"1350\" fill=\"#f4eee3\"\u002F>\n\u003Cg transform=\"translate(60 62) scale(1.375)\">\u003Crect x=\"2\" y=\"17\" width=\"6\" height=\"13\" rx=\"1.2\" fill=\"#0e6f97\"\u002F>\u003Crect x=\"13\" y=\"9\" width=\"6\" height=\"21\" rx=\"1.2\" fill=\"#3e7a1d\"\u002F>\u003Crect x=\"24\" y=\"2\" width=\"6\" height=\"28\" rx=\"1.2\" fill=\"#9b7424\"\u002F>\u003C\u002Fg>\n\u003Ctext x=\"118\" y=\"99\" font-family=\"'Fraunces Variable', Georgia, serif\" font-weight=\"600\" font-size=\"36\" fill=\"#1b1915\">Infografx\u003C\u002Ftext>\n\u003Ctext x=\"1020\" y=\"96\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"24\" letter-spacing=\"3\" fill=\"#0e6f97\" text-anchor=\"end\">AI\u003C\u002Ftext>\n\u003Crect x=\"60\" y=\"132\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\n\u003Ctext font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"26\" letter-spacing=\"2\" fill=\"#6e675d\">\u003Ctspan x=\"60\" y=\"192\">HOW OFTEN DID GPT-6.2 ASTRA ANSWER CORRECTLY ON ALL\u003C\u002Ftspan>\u003Ctspan x=\"60\" y=\"230\">FIVE ATTEMPTS?\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext font-family=\"'Fraunces Variable', Georgia, serif\" font-weight=\"600\" font-size=\"68\" letter-spacing=\"-0.5\" fill=\"#1b1915\">\u003Ctspan x=\"60\" y=\"322\">62.5% right all five times\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Crect x=\"60\" y=\"1070\" width=\"4\" height=\"84\" fill=\"#9b7424\"\u002F>\n\u003Ctext font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"30\" fill=\"#45403a\">\u003Ctspan x=\"88\" y=\"1100\">The test measures how often answers stay correct\u003C\u002Ftspan>\u003Ctspan x=\"88\" y=\"1142\">across repeated attempts.\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"22\" fill=\"#6e675d\">\u003Ctspan x=\"60\" y=\"1204\">Source: Artificial Analysis — AA-AnalystAgent Benchmark · Leaderboard\u003C\u002Ftspan>\u003Ctspan x=\"60\" y=\"1236\">as of 19 November 2026 · Unit: percent\u003C\u002Ftspan>\u003C\u002Ftext>\n\u003Ctext x=\"60\" y=\"396\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"22\" fill=\"#6e675d\">GPT-6.2 ASTRA · PASS^5 SCORE BY SETTING, %\u003C\u002Ftext>\u003Crect x=\"199\" y=\"531\" width=\"2\" height=\"372\" fill=\"#d9cfbe\"\u002F>\u003Ctext x=\"60\" y=\"598\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#1b1915\">Max\u003C\u002Ftext>\u003Crect x=\"200\" y=\"555\" width=\"640.0\" height=\"64\" rx=\"3\" fill=\"#0e6f97\" data-label=\"Max\" data-value=\"62.5\"\u002F>\u003Ctext x=\"856.0\" y=\"599.24\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"34\" fill=\"#0e6f97\">62.5%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"728\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#1b1915\">Xhigh\u003C\u002Ftext>\u003Crect x=\"200\" y=\"685\" width=\"640.0\" height=\"64\" rx=\"3\" fill=\"#0e6f97\" data-label=\"Xhigh\" data-value=\"62.5\"\u002F>\u003Ctext x=\"856.0\" y=\"729.24\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"34\" fill=\"#0e6f97\">62.5%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"858\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"30\" fill=\"#45403a\">High\u003C\u002Ftext>\u003Crect x=\"200\" y=\"815\" width=\"627.7\" height=\"64\" rx=\"3\" fill=\"#cfc4b1\" data-label=\"High\" data-value=\"61.3\"\u002F>\u003Ctext x=\"843.7\" y=\"859.24\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"34\" fill=\"#1b1915\">61.3%\u003C\u002Ftext>\u003Ctext x=\"60\" y=\"1028\" font-family=\"'Inter Variable', system-ui, sans-serif\" font-size=\"24\" fill=\"#6e675d\">A question counts only if all five attempts are correct.\u003C\u002Ftext>\n\u003Crect x=\"0\" y=\"1266\" width=\"1080\" height=\"84\" fill=\"#f4eee3\"\u002F>\n\u003Crect x=\"60\" y=\"1266\" width=\"960\" height=\"1\" fill=\"#d9cfbe\"\u002F>\n\u003Cg transform=\"translate(60 1282) scale(0.9375)\">\u003Crect x=\"2\" y=\"17\" width=\"6\" height=\"13\" rx=\"1.2\" fill=\"#0e6f97\"\u002F>\u003Crect x=\"13\" y=\"9\" width=\"6\" height=\"21\" rx=\"1.2\" fill=\"#3e7a1d\"\u002F>\u003Crect x=\"24\" y=\"2\" width=\"6\" height=\"28\" rx=\"1.2\" fill=\"#9b7424\"\u002F>\u003C\u002Fg>\n\u003Ctext x=\"104\" y=\"1307\" font-family=\"'JetBrains Mono', ui-monospace, monospace\" font-weight=\"500\" font-size=\"26\" fill=\"#1b1915\" fill-opacity=\"0.7\">infografx.org\u003C\u002Ftext>\n\u003C\u002Fsvg>\n",1795075269750]