Home Model series Claude Opus Claude Opus · LAB-Bench FigQA · FutureHouse FigQA; prompting and tools are recorded per result · Release history Claude Opus · LAB-Bench FigQA · FutureHouse FigQA; prompting and tools are recorded per result · Release history View scores, configurations, and disclosures for LAB-Bench FigQA · FutureHouse FigQA; prompting and tools are recorded per result across 4 Claude Opus releases, ordered by release date.
Differences are calculated only for unique, active, exact values between direct predecessors when the unit, metric, and explicitly recorded configuration and conditions match.
2025-11-24 · 2 score records for this release, including 2 exact values (may include historical records)
Claude Opus 4.5 · LAB-Bench FigQA scores and configurations Metric and score Configuration and evaluation conditions Sources Score 54.9% Reported 54.9% · Unit % Published Unconfirmed
LAB-Bench FigQA, no tools Settings Extended thinking off; no tools; zero-shot prompting. Unknowns Trial count is not stated. Tools Tools forbidden Reasoning budget not recorded Framework not recorded Protocol LAB-Bench FigQA; zero-shot prompting; 95% confidence interval shown in the figure. Conditions No tools; extended thinking off. View this score in rankings Score 69.2% Reported 69.2% · Unit % Published Unconfirmed
LAB-Bench FigQA, image cropping tool Settings Simple image cropping tool; 32,768 reasoning-token budget; zero-shot prompting. Unknowns Trial count is not stated. Tools Tools allowed Reasoning budget Thinking budget 32,768 Framework not recorded Protocol LAB-Bench FigQA; zero-shot prompting; 95% confidence interval shown in the figure. Conditions Simple image cropping tool; 32,768 reasoning-token budget. View this score in rankings
PDF pp.33-34, §2.21, Figure 2.21.A Attachment · Claude Opus 4.5 System Card 2026-02-05 · 2 score records for this release, including 2 exact values (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Claude Opus 4.6 · LAB-Bench FigQA scores and configurations Metric and score Configuration and evaluation conditions Sources Score 58% Reported 58.0% · Unit % Published Unconfirmed
Max · 无工具 Settings Adaptive thinking、Max effort、无工具;sampling 使用默认 temperature 与 top_p。 Unknowns none recorded Tools Tools forbidden Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA;without tools。 Conditions Adaptive thinking、max effort、5-run average。 View this score in rankings Score 78.3% Reported 78.3% · Unit % Published Unconfirmed
Max · 图像裁切工具 Settings Adaptive thinking、Max effort;允许简单 image-cropping tool;每项评测条件见成绩记录。 Unknowns none recorded Tools Tools allowed Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA;simple image-cropping tool。 Conditions Adaptive thinking、max effort、5-run average。 View this score in rankings
PDF pp.34–35, §2.19.1 and Figure 2.19.1.A Attachment · Claude Opus 4.6 System Card 2026-04-16 · 4 score records for this release, including 4 exact values (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Claude Opus 4.7 · LAB-Bench FigQA scores and configurations Metric and score Configuration and evaluation conditions Sources Score 78.6% Reported 78.6% · Unit % Published Unconfirmed
LAB-Bench FigQA,无工具 Settings Adaptive thinking, max effort; no tools; 5-run average; current maximum image resolution (2576px single dimension, 3.75MP total). Unknowns none recorded Tools Tools forbidden Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA; no tools. Conditions Adaptive thinking, max effort; five-run average; current image limit 2576px on one dimension and 3.75MP total. View this score in rankings Score 86.4% Reported 86.4% · Unit % Published Unconfirmed
LAB-Bench FigQA,Python工具 Settings Adaptive thinking, max effort; Python tools; 5-run average; current maximum image resolution (2576px single dimension, 3.75MP total). Unknowns none recorded Tools Tools allowed Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA; Python tools. Conditions Adaptive thinking, max effort; five-run average; current image limit 2576px on one dimension and 3.75MP total. View this score in rankings Score 74% Reported 74.0% · Unit % Published Unconfirmed
LAB-Bench FigQA,无工具,旧分辨率 Settings Adaptive thinking, max effort; no tools; 5-run average; old image limit (1568px single dimension, 1.15MP total). Unknowns none recorded Tools Tools forbidden Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA; no tools; adaptive thinking; max effort; old image resolution. Conditions Five-run average; image maximum 1568px on one dimension and 1.15MP total. Separate resolution ablation, not the current-resolution 78.6% result. View this score in rankings Score 85.8% Reported 85.8% · Unit % Published Unconfirmed
LAB-Bench FigQA,Python工具,旧分辨率 Settings Adaptive thinking, max effort; Python tools; 5-run average; old image limit (1568px single dimension, 1.15MP total). Unknowns none recorded Tools Tools allowed Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA; Python tools; adaptive thinking; max effort; old image resolution. Conditions Five-run average; image maximum 1568px on one dimension and 1.15MP total. Separate resolution ablation, not the current-resolution 86.4% result. View this score in rankings
§8.9.1, PDF pp.203-204 Attachment · Claude Opus 4.7 System Card §8.9.1 and Figure 8.9.1.B, PDF pp.203-204 Attachment · Claude Opus 4.7 System Card 2026-05-28 · 2 score records for this release, including 2 exact values (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Claude Opus 4.8 · LAB-Bench FigQA scores and configurations Metric and score Configuration and evaluation conditions Sources Score 80.4% Reported 80.4% · Unit % Published Unconfirmed
视觉基准无工具 Settings Adaptive thinking、max effort、无工具;各图表注明时平均 5 runs,95% CI。 Unknowns none recorded Tools Tools forbidden Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA public set, 181 questions; average over 5 runs; figure uses new grader. Conditions Adaptive thinking/max effort, no tools. Proposed catalog ID; do not merge with LABBench2 overall/category score. View this score in rankings Score 87.3% Reported 87.3% · Unit % Published Unconfirmed
视觉基准 Python 工具 Settings Adaptive thinking、max effort、Python tools;各图表注明时平均 5 runs,95% CI。 Unknowns none recorded Tools Tools allowed Reasoning budget max Framework not recorded Protocol LAB-Bench FigQA public set, 181 questions; average over 5 runs; figure uses new grader. Conditions Adaptive thinking/max effort with Python tools. Proposed catalog ID; do not merge with LABBench2 overall/category score. View this score in rankings