{"id":19748,"date":"2026-09-24T05:09:31","date_gmt":"2026-09-24T09:09:31","guid":{"rendered":"https:\/\/wp.glbgpt.com\/?p=19748"},"modified":"2026-09-24T05:09:32","modified_gmt":"2026-09-24T09:09:32","slug":"claude-opus-5-5-vs-opus-5-vs-fable-5-1","status":"publish","type":"post","link":"https:\/\/wp.glbgpt.com\/pt-br\/hub\/claude-opus-5-5-vs-opus-5-vs-fable-5-1","title":{"rendered":"Claude Opus 5.5 x Opus 5 x Fable 5.1: Testes reais"},"content":{"rendered":"<p class=\"wp-block-paragraph\">Anthropic\u2019s current Claude lineup makes a simple price comparison misleading. This review puts Claude Opus 5.5, Claude Opus 5, and Claude Fable 5.1 through the same five prompts and reports the task result, local elapsed time, route-reported token usage, and the limits of each observation.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>The practical difference is visible in the task cards.<\/strong> Opus 5.5 was the quickest and cheapest on the measured pack, Fable 5.1 produced the fewest output tokens and the cleanest Chinese-format response, and Opus 5 took the most time, while Opus 5.5 used the most output tokens. The coding card adds an important qualification: shorter output was not automatically better output.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">We copied the five prompts from the earlier Claude Opus 5.5 review and ran every prompt once for each model. The request settings were identical: an 8,192-token ceiling, high effort, one user message, and no tools. The route returned HTTP 200 and <code>end_turn<\/code> for all 15 requests.<\/p>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-quick\" id=\"quick-answer\" style=\"margin:24px 0;padding:22px 25px;border:1px solid #dfe5ef;border-radius:16px;background:#f4f2ff;border-left:6px solid #635bdb;\"><h2 style=\"margin:54px 0 14px;padding-left:14px;border-left:5px solid #635bdb;font-size:30px;line-height:1.2;margin:0 0 10px;padding:0;border:0;font-size:24px;\">Resposta r\u00e1pida<\/h2><ul><li style=\"margin:7px 0;\"><strong>Fastest in this five-task pack:<\/strong> Claude Opus 5.5 used 47.725 seconds of local end-to-end time across five sequential requests. Claude Fable 5.1 used 53.290 seconds, and Claude Opus 5 used 61.470 seconds.<\/li><li style=\"margin:7px 0;\"><strong>Lowest list-price estimate:<\/strong> Opus 5.5 came to about $0.0818 for the five requests at official token rates, followed by Opus 5 at $0.0962 and Fable 5.1 at $0.1415. These are arithmetic estimates, not platform charges.<\/li><li style=\"margin:7px 0;\"><strong>Quality result:<\/strong> all three returned grounded five-bullet extraction, parseable JSON, and the correct math answer. The coding outputs differed in edge-case handling, while Fable 5.1 followed the two-paragraph Chinese-format instruction most cleanly.<\/li><li style=\"margin:7px 0;\"><strong>API caveat:<\/strong> Opus 5.5 is not a drop-in rename for every Opus 5 integration. Its migration guide says thinking cannot be disabled and some older tool\/thinking patterns need changes.<\/li><li style=\"margin:7px 0;\"><strong>Do not overread the ranking:<\/strong> this is one matched run per task on one route. It measures concrete behavior, elapsed time, and reported usage, not a universal model score.<\/li><\/ul><\/section><\/div>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"specs\">What the three models cost and expose<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The three APIs share a 1M-token context window and a 128K maximum output in Anthropic\u2019s current model documentation. Their billing units and status differ: Opus 5.5 is the latest Opus route, Opus 5 is marked legacy, and Fable 5.1 carries the highest standard input\/output rate.<\/p>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Status oficial<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">ID do modelo da API<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Entrada \/ sa\u00edda<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Contexto \/ pot\u00eancia m\u00e1xima<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Default effort<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Leitura do cache<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Mais recentes<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><code style=\"padding:2px 6px;border-radius:5px;background:#eef1f7;color:#293657;\">claude-opus-5-5<\/code><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">$4 \/ $20 per MTok<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">1M \/ 128K<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">M\u00e9dio<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">$0.20 \/ MTok<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Active, legacy<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><code style=\"padding:2px 6px;border-radius:5px;background:#eef1f7;color:#293657;\">claude-opus-5<\/code><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">$5 \/ $25 por MTok<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">1M \/ 128K<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Alto<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">$0.50 \/ MTok<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Mais recentes<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><code style=\"padding:2px 6px;border-radius:5px;background:#eef1f7;color:#293657;\">claude-f\u00e1bula-5-1<\/code><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">$10 \/ $50 por MTok<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">1M \/ 128K<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Alto<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">$0.25 \/ MTok<\/td><\/tr><\/tbody><\/table><\/div><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">For a simple non-cached example, 100,000 input tokens plus 20,000 output tokens costs $0.80 on Opus 5.5, $1.00 on Opus 5, and $2.00 on Fable 5.1 at the listed rates. Fable 5.1 has a lower cache-read rate than Opus 5, so cache-heavy agent workloads need a separate calculation.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"official-performance\">What the official benchmark table shows<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Anthropic\u2019s <a href=\"https:\/\/www.anthropic.com\/claude-opus-5-5\">Claude Opus 5.5 announcement<\/a> places all three models in the same provider comparison table. Opus 5.5 leads the listed rows, but these are provider-reported results with benchmark-specific harnesses and effort settings. The announcement itself warns that narrow benchmark margins are a less reliable guide to real-world differences at this capability level.<\/p>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Provider benchmark<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Opus 5.5<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Fable 5.1<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Opus 5<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">What it measures<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Terminal-Bench 4.0<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>66.4%<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">55.8%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">52.3%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Agentic command-line coding<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">FrontierCode v1.1<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>54.4%<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">50.3%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">48.0%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Whether agent code changes would be merged<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">CursorBench 4.0<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>57.8%<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">51.8%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">46.6%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Ambiguous multi-file coding tasks<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">GDPval-AA v2.1<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>1846<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">1735<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">1708<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Professional knowledge work across 44 occupations<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">AutomationBench<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>40.0%<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">31.4%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">26.9%<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">End-to-end business workflows<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">O \u00faltimo exame da humanidade<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>67.7%<\/strong> with tools<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">65.6% with tools<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">63.6% with tools<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Multidisciplinary reasoning<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">OSWorld 2.0<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>81.8%<\/strong> partial<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">80.7% partial<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">74.0% partial<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Computer-use tasks<\/td><\/tr><\/tbody><\/table><\/div><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">The table is useful for locating the provider\u2019s claimed performance frontier. It is not a replacement for the matched cards below: our test uses short single-turn prompts, high effort, no tools, and a third-party controlled route.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"matched-test\">Same prompts, same request settings<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">We copied the five prompts from the earlier Opus 5.5 test pack and sent each one once to each model. The client rotated model order by task, used UTF-8 JSON, and measured from request start until the complete response body was read. All 15 requests returned HTTP 200 and <code>end_turn<\/code>. For the broader rationale and repeatable workflow, see <a href=\"https:\/\/www.glbgpt.com\/hub\/how-globalgpt-tests-ai-models\/\">How GlobalGPT tests AI models<\/a>.<\/p>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-method\" style=\"margin:24px 0;padding:22px 25px;border:1px solid #dfe5ef;border-radius:16px;background:#f4f2ff;background:#eef6ff;border-color:#cbdcf8;\"><strong>Measurement boundary:<\/strong> the times below are local end-to-end wall-clock measurements. They include the route and network path used for this test, so they are not provider latency. Token counts and the thinking field are values reported by the route. The complete raw responses, prompt hashes, and run metadata are kept in the editor package.<\/div><\/div>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"aggregate\">Time and token comparison<\/h2>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Rota<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">5-run elapsed<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Mean \/ task<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Tokens de entrada<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Tokens de sa\u00edda<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Route-reported thinking<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">List-price estimate*<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">47.725s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">9.545s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">700<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">3,952<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">2,214<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>$0.0818<\/strong><\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">61.470s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">12.294s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">690<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">3,709<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">1,859<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">$0.0962<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">53.290s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">10.658s<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">700<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">2,690<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">1,122<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">$0.1415<\/td><\/tr><\/tbody><\/table><\/div><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">*Estimate from route-reported input\/output tokens and Anthropic\u2019s listed standard API rates. It excludes cache charges, platform credits, taxes, and markup. Elapsed time is local end-to-end wall time after the request was sent, not provider latency or tokens per second. \u201cThinking\u201d is a route-reported usage field, not independently measured hidden reasoning.<\/p>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-grid\" style=\"display:grid;grid-template-columns:repeat(3,minmax(0,1fr));gap:14px;margin:24px 0;grid-template-columns:repeat(auto-fit,minmax(210px,1fr));\"><div class=\"cc-stat\" style=\"padding:18px;border:1px solid #dfe5ef;border-radius:14px;background:#fbfcff;\"><strong style=\"display:block;font-size:28px;color:#635bdb;\">47.725s<\/strong><span style=\"display:block;color:#64748b;font-size:13px;\">Opus 5.5 total local time<\/span><\/div><div class=\"cc-stat\" style=\"padding:18px;border:1px solid #dfe5ef;border-radius:14px;background:#fbfcff;\"><strong style=\"display:block;font-size:28px;color:#635bdb;\">2,690<\/strong><span style=\"display:block;color:#64748b;font-size:13px;\">Fable 5.1 output tokens<\/span><\/div><div class=\"cc-stat\" style=\"padding:18px;border:1px solid #dfe5ef;border-radius:14px;background:#fbfcff;\"><strong style=\"display:block;font-size:28px;color:#635bdb;\">$0.0818<\/strong><span style=\"display:block;color:#64748b;font-size:13px;\">Opus 5.5 five-run list-price estimate<\/span><\/div><\/div><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">The aggregate points in different directions depending on the metric. Opus 5.5 returned the fastest total and the lowest estimated cost, Fable 5.1 returned the fewest output tokens, and Opus 5 had the slowest total in this pack. A shorter answer is not automatically a better answer: the coding card shows why output content and edge-case handling still need to be read.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"task-cards\">Five matched task cards<\/h2>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-task\" id=\"coding-debug\" style=\"margin:28px 0;padding:22px;border:1px solid #dfe5ef;border-radius:18px;background:#fff;box-shadow:0 10px 26px #10182812;\"><div class=\"cc-task-head\" style=\"display:flex;justify-content:space-between;gap:16px;align-items:flex-start;margin-bottom:14px;\"><div><span class=\"cc-task-kicker\" style=\"color:#5c55bd;font-size:11px;font-weight:800;letter-spacing:.08em;text-transform:uppercase;\">Tarefa 01 \u00b7 Depura\u00e7\u00e3o em Python<\/span><h3 style=\"line-height:1.25;margin:3px 0;font-size:23px;\">Can the models repair a mixed-type deduplication function?<\/h3><\/div><span class=\"cc-pill\" style=\"display:inline-flex;padding:5px 10px;border-radius:999px;background:#eef3ff;color:#334c9f;font-size:12px;font-weight:800;\">Matched prompt \u00b7 one run each<\/span><\/div><p style=\"margin:0 0 18px;\"><strong>Configura\u00e7\u00e3o da tarefa:<\/strong> The prompt required a corrected function plus two concise tests, case-insensitive string comparison, preserved first spelling, and no crash on non-string values.<\/p><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Time \/ usage<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Resultado observado<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">15.844s local \u00b7 145 in \/ 1478 out \u00b7 700 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Uses casefold plus a list fallback; its docstring says identity while the code uses equality.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">16.845s local \u00b7 143 in \/ 1109 out \u00b7 312 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Adds type-tagged keys and an equality fallback, and calls out the invalid one-line Python syntax.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">12.778s local \u00b7 145 in \/ 783 out \u00b7 0 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Uses repr(item) for unhashable values; that passes the shown examples but can collide across types.<\/td><\/tr><\/tbody><\/table><\/div><p class=\"cc-observation\" style=\"margin:0 0 18px;margin:14px 0;padding:12px 14px;border-left:4px solid #635bdb;background:#f5f3ff;\"><strong>Comparison result:<\/strong> The edge case is a task-level observation, not a claim about all coding workloads.<\/p><\/section><\/div>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-task\" id=\"long-context-extract\" style=\"margin:28px 0;padding:22px;border:1px solid #dfe5ef;border-radius:18px;background:#fff;box-shadow:0 10px 26px #10182812;\"><div class=\"cc-task-head\" style=\"display:flex;justify-content:space-between;gap:16px;align-items:flex-start;margin-bottom:14px;\"><div><span class=\"cc-task-kicker\" style=\"color:#5c55bd;font-size:11px;font-weight:800;letter-spacing:.08em;text-transform:uppercase;\">Task 02 \u00b7 Grounded extraction<\/span><h3 style=\"line-height:1.25;margin:3px 0;font-size:23px;\">Can the models preserve five supplied facts without adding claims?<\/h3><\/div><span class=\"cc-pill\" style=\"display:inline-flex;padding:5px 10px;border-radius:999px;background:#eef3ff;color:#334c9f;font-size:12px;font-weight:800;\">Matched prompt \u00b7 one run each<\/span><\/div><p style=\"margin:0 0 18px;\"><strong>Configura\u00e7\u00e3o da tarefa:<\/strong> The input was a short note about a two-week test of three writing assistants. It is a short extraction task, not a large-context benchmark.<\/p><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Time \/ usage<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Resultado observado<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">4.374s local \u00b7 210 in \/ 312 out \u00b7 101 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned five numbered bullets and stayed inside the supplied facts; labels and order differed slightly.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">4.661s local \u00b7 208 in \/ 265 out \u00b7 27 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Returned five numbered bullets and stayed inside the supplied facts; labels and order differed slightly.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">6.305s local \u00b7 210 in \/ 223 out \u00b7 0 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned five numbered bullets and stayed inside the supplied facts; labels and order differed slightly.<\/td><\/tr><\/tbody><\/table><\/div><p class=\"cc-observation\" style=\"margin:0 0 18px;margin:14px 0;padding:12px 14px;border-left:4px solid #635bdb;background:#f5f3ff;\"><strong>Comparison result:<\/strong> The source passage was short; this checks grounded extraction and formatting, not million-token context handling.<\/p><\/section><\/div>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-task\" id=\"structured-json\" style=\"margin:28px 0;padding:22px;border:1px solid #dfe5ef;border-radius:18px;background:#fff;box-shadow:0 10px 26px #10182812;\"><div class=\"cc-task-head\" style=\"display:flex;justify-content:space-between;gap:16px;align-items:flex-start;margin-bottom:14px;\"><div><span class=\"cc-task-kicker\" style=\"color:#5c55bd;font-size:11px;font-weight:800;letter-spacing:.08em;text-transform:uppercase;\">Task 03 \u00b7 JSON compliance<\/span><h3 style=\"line-height:1.25;margin:3px 0;font-size:23px;\">Can the models return the exact requested structure?<\/h3><\/div><span class=\"cc-pill\" style=\"display:inline-flex;padding:5px 10px;border-radius:999px;background:#eef3ff;color:#334c9f;font-size:12px;font-weight:800;\">Matched prompt \u00b7 one run each<\/span><\/div><p style=\"margin:0 0 18px;\"><strong>Configura\u00e7\u00e3o da tarefa:<\/strong> The prompt required valid JSON only, five exact keys, two pros, two cons, and short strings. The fictional review content is not evidence about any model.<\/p><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Time \/ usage<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Resultado observado<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">8.276s local \u00b7 128 in \/ 622 out \u00b7 390 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned parseable JSON with the requested keys and two pros\/two cons; generated review claims are test text, not product facts.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">6.344s local \u00b7 126 in \/ 238 out \u00b7 0 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Returned parseable JSON with the requested keys and two pros\/two cons; generated review claims are test text, not product facts.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">6.626s local \u00b7 128 in \/ 203 out \u00b7 0 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned parseable JSON with the requested keys and two pros\/two cons; generated review claims are test text, not product facts.<\/td><\/tr><\/tbody><\/table><\/div><p class=\"cc-observation\" style=\"margin:0 0 18px;margin:14px 0;padding:12px 14px;border-left:4px solid #635bdb;background:#f5f3ff;\"><strong>Comparison result:<\/strong> The JSON strings describe a fictional review setup and must not be published as product facts.<\/p><\/section><\/div>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-task\" id=\"math-reasoning\" style=\"margin:28px 0;padding:22px;border:1px solid #dfe5ef;border-radius:18px;background:#fff;box-shadow:0 10px 26px #10182812;\"><div class=\"cc-task-head\" style=\"display:flex;justify-content:space-between;gap:16px;align-items:flex-start;margin-bottom:14px;\"><div><span class=\"cc-task-kicker\" style=\"color:#5c55bd;font-size:11px;font-weight:800;letter-spacing:.08em;text-transform:uppercase;\">Task 04 \u00b7 Arithmetic<\/span><h3 style=\"line-height:1.25;margin:3px 0;font-size:23px;\">Can the models carry the rounded batch sequence through to the final answer?<\/h3><\/div><span class=\"cc-pill\" style=\"display:inline-flex;padding:5px 10px;border-radius:999px;background:#eef3ff;color:#334c9f;font-size:12px;font-weight:800;\">Matched prompt \u00b7 one run each<\/span><\/div><p style=\"margin:0 0 18px;\"><strong>Configura\u00e7\u00e3o da tarefa:<\/strong> The sequence was 12, then 25% more with floor rounding, across four batches.<\/p><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Time \/ usage<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Resultado observado<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">3.830s local \u00b7 89 in \/ 257 out \u00b7 74 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Computed 12, 15, 18, 22 and a total of 67; all three ended with the requested standalone 67.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">5.079s local \u00b7 87 in \/ 277 out \u00b7 90 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Computed 12, 15, 18, 22 and a total of 67; all three ended with the requested standalone 67.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">5.347s local \u00b7 89 in \/ 127 out \u00b7 0 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Computed 12, 15, 18, 22 and a total of 67; all three ended with the requested standalone 67.<\/td><\/tr><\/tbody><\/table><\/div><p class=\"cc-observation\" style=\"margin:0 0 18px;margin:14px 0;padding:12px 14px;border-left:4px solid #635bdb;background:#f5f3ff;\"><strong>Comparison result:<\/strong> All three passed this arithmetic prompt; one prompt cannot estimate general reasoning reliability.<\/p><\/section><\/div>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><section class=\"cc-task\" id=\"chinese-seo-outline\" style=\"margin:28px 0;padding:22px;border:1px solid #dfe5ef;border-radius:18px;background:#fff;box-shadow:0 10px 26px #10182812;\"><div class=\"cc-task-head\" style=\"display:flex;justify-content:space-between;gap:16px;align-items:flex-start;margin-bottom:14px;\"><div><span class=\"cc-task-kicker\" style=\"color:#5c55bd;font-size:11px;font-weight:800;letter-spacing:.08em;text-transform:uppercase;\">Task 05 \u00b7 Chinese format following<\/span><h3 style=\"line-height:1.25;margin:3px 0;font-size:23px;\">Can the route keep the requested two-paragraph structure?<\/h3><\/div><span class=\"cc-pill\" style=\"display:inline-flex;padding:5px 10px;border-radius:999px;background:#eef3ff;color:#334c9f;font-size:12px;font-weight:800;\">Matched prompt \u00b7 one run each<\/span><\/div><p style=\"margin:0 0 18px;\"><strong>Configura\u00e7\u00e3o da tarefa:<\/strong> The prompt requested two Chinese paragraphs of 80\u2013120 Chinese characters each: a value answer followed by the test plan.<\/p><div class=\"cc-table-wrap\" style=\"overflow-x:auto;margin:22px 0;\"><table class=\"cc-table\" style=\"width:100%;border-collapse:collapse;min-width:700px;width:100%;border-collapse:collapse;min-width:650px;\"><thead><tr><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Modelo<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Time \/ usage<\/th><th style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#f1f3fa;font-size:13px;\">Resultado observado<\/th><\/tr><\/thead><tbody><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Opus 5.5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">15.401s local \u00b7 128 in \/ 1283 out \u00b7 949 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned the two requested paragraphs, then added Markdown framing and an English note.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\"><strong>Claude Opus 5<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">28.541s local \u00b7 126 in \/ 1820 out \u00b7 1430 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;background:#fbfcff;\">Returned the two requested paragraphs, then added an English note after a separator.<\/td><\/tr><tr><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\"><strong>Claude Fable 5.1<\/strong><\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">22.234s local \u00b7 128 in \/ 1354 out \u00b7 1122 thinking \u00b7 end_turn<\/td><td style=\"padding:12px 13px;border:1px solid #dfe5ef;text-align:left;vertical-align:top;\">Returned two clean Chinese paragraphs with no extra note or Markdown framing.<\/td><\/tr><\/tbody><\/table><\/div><p class=\"cc-observation\" style=\"margin:0 0 18px;margin:14px 0;padding:12px 14px;border-left:4px solid #635bdb;background:#f5f3ff;\"><strong>Comparison result:<\/strong> This records instruction following through the tested route; it is not a Chinese-language quality benchmark.<\/p><\/section><\/div>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"api-caveats\">Considera\u00e7\u00f5es sobre API e migra\u00e7\u00e3o<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Opus 5.5 is not a drop-in rename for every Opus 5 integration. Anthropic\u2019s <a href=\"https:\/\/platform.claude.com\/docs\/en\/models\/opus-5-5\/migration-guide\">migration guide<\/a> says thinking cannot be disabled on Opus 5.5; requests that try to disable or manually enable the old thinking mode return a 400 error. The migration also changes forced tool use, ties thinking blocks to the model and conversation, and replaces the older computer-use tool on the Claude API and Google Cloud path.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">The shared test used <code>output_config.effort: high<\/code> for all three models so the prompt results were easier to compare. That means these runs do not reproduce each model\u2019s default effort setting. Fable 5.1 is documented with adaptive thinking always on and high default effort; Opus 5.5 is documented with adaptive thinking always on and medium default effort; Opus 5 uses adaptive thinking with high default effort.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"verdict\">What this test actually supports<\/h2>\n\n\n\n<div class=\"claude-compare\" style=\"color:#172033;font:16px\/1.72 Inter,ui-sans-serif,-apple-system,BlinkMacSystemFont,Segoe UI,Arial,sans-serif;\"><div class=\"cc-verdict\" style=\"margin:24px 0;padding:22px 25px;border:1px solid #dfe5ef;border-radius:16px;background:#f4f2ff;background:#111a36;color:#eff3ff;border:0;\"><h2 style=\"margin:54px 0 14px;padding-left:14px;border-left:5px solid #635bdb;font-size:30px;line-height:1.2;margin:0 0 10px;padding:0;border:0;font-size:24px;color:#fff;\">Task-level verdict<\/h2><p style=\"margin:0 0 18px;color:#e8edff;\"><strong>Opus 5.5:<\/strong> fastest and lowest estimated list-price cost in this five-task pack, while also producing the most output tokens. <strong>Fable 5.1:<\/strong> shortest output overall and cleanest compliance with the Chinese two-paragraph constraint, but its higher token price made the five-run estimate highest and its coding fallback had a concrete cross-type edge case. <strong>Opus 5:<\/strong> correct on the extraction, JSON, and math tasks, but slowest in this route sample and marked legacy in the official docs.<\/p><p style=\"margin:0 0 18px;color:#e8edff;\">That is enough to justify a practical preference for Opus 5.5 on this measured route, not enough to claim that it wins every workload. Repeat the test with tools, longer inputs, and multiple runs before treating the result as a production benchmark.<\/p><\/div><\/div>\n\n\n\n<p class=\"wp-block-paragraph\">If you want to reproduce the comparison without opening three separate accounts, <a href=\"https:\/\/www.glbgpt.com\/home?inviter=hub_popup&amp;login=1\">try the model comparison workflow in GlobalGPT<\/a>. The direct model route and the upstream provider mapping should still be verified separately for any production decision.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"faq\">PERGUNTAS FREQUENTES<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Which model was fastest in the matched test?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Claude Opus 5.5 had the lowest total local elapsed time: 47.725 seconds across five sequential requests. This is a route observation, not provider-side latency.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Which model used the fewest output tokens?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Claude Fable 5.1 used 2,690 route-reported output tokens across the five tasks, compared with 3,709 for Opus 5 and 3,952 for Opus 5.5.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Which model was cheapest in the test?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Using official standard API rates and the returned input\/output token counts, Opus 5.5 produced the lowest five-run estimate at $0.0818. This excludes cache, credits, taxes, and platform markup.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Did all three models pass the same tasks?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">All three returned grounded five-bullet extraction, parseable JSON with the requested keys, and the correct total of 67. The coding and Chinese-format cards show meaningful output differences.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Was this a long-context benchmark?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">No. The reused extraction prompt contains a short passage. It tests grounding and exact formatting, so it should not be presented as evidence of million-token context performance.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Is Claude Opus 5 still current?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Anthropic\u2019s model docs label Opus 5 as active but legacy and recommend considering Opus 5.5 for improved performance. Existing API integrations should review the migration guide before switching.<\/p>\n\n\n\n<script type=\"application\/ld+json\">{\n    \"@context\": \"https:\\\/\\\/schema.org\",\n    \"@type\": \"FAQPage\",\n    \"mainEntity\": [\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Which model was fastest in the matched test?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Claude Opus 5.5 had the lowest total local elapsed time: 47.725 seconds across five sequential requests. This is a route observation, not provider-side latency.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Which model used the fewest output tokens?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Claude Fable 5.1 used 2,690 route-reported output tokens across the five tasks, compared with 3,709 for Opus 5 and 3,952 for Opus 5.5.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Which model was cheapest in the test?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Using official standard API rates and the returned input\\\/output token counts, Opus 5.5 produced the lowest five-run estimate at $0.0818. This excludes cache, credits, taxes, and platform markup.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Did all three models pass the same tasks?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"All three returned grounded five-bullet extraction, parseable JSON with the requested keys, and the correct total of 67. The coding and Chinese-format cards show meaningful output differences.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Was this a long-context benchmark?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"No. The reused extraction prompt contains a short passage. It tests grounding and exact formatting, so it should not be presented as evidence of million-token context performance.\"\n            }\n        },\n        {\n            \"@type\": \"Question\",\n            \"name\": \"Is Claude Opus 5 still current?\",\n            \"acceptedAnswer\": {\n                \"@type\": \"Answer\",\n                \"text\": \"Anthropic\\u2019s model docs label Opus 5 as active but legacy and recommend considering Opus 5.5 for improved performance. Existing API integrations should review the migration guide before switching.\"\n            }\n        }\n    ]\n}<\/script>\n\n\n\n<p class=\"wp-block-paragraph\">Official specifications: <a href=\"https:\/\/platform.claude.com\/docs\/en\/models\/opus-5-5\/overview\">Opus 5.5 docs<\/a>, <a href=\"https:\/\/platform.claude.com\/docs\/en\/models\/opus-5\/overview\">Opus 5 docs<\/a>, e <a href=\"https:\/\/platform.claude.com\/docs\/en\/models\/fable-5-1\/overview\">Fable 5.1 docs<\/a>. Provider benchmark values are attributed to Anthropic; hands-on values come from the matched test pack described above.<\/p>","protected":false},"excerpt":{"rendered":"<p>Claude Opus 5.5 vs Opus 5 vs Fable 5.1: compare pre\u00e7os oficiais, pontua\u00e7\u00f5es em benchmarks, tempo de resposta local, uso de tokens e cinco tarefas de API com o mesmo prompt.<\/p>","protected":false},"author":13,"featured_media":19754,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_seopress_robots_primary_cat":"","_seopress_titles_title":"Claude Opus 5.5 vs Opus 5 vs Fable 5.1: Real Tests","_seopress_titles_desc":"Claude Opus 5.5 vs Opus 5 vs Fable 5.1: compare official prices, benchmark scores, local response time, token use, and five same-prompt API tasks.","_seopress_robots_index":"","footnotes":""},"categories":[7],"tags":[],"class_list":["post-19748","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-chat"],"acf":[],"_links":{"self":[{"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/posts\/19748","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/users\/13"}],"replies":[{"embeddable":true,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/comments?post=19748"}],"version-history":[{"count":1,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/posts\/19748\/revisions"}],"predecessor-version":[{"id":19755,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/posts\/19748\/revisions\/19755"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/media\/19754"}],"wp:attachment":[{"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/media?parent=19748"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/categories?post=19748"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/wp.glbgpt.com\/pt-br\/wp-json\/wp\/v2\/tags?post=19748"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}