75.8%
ARC-AGI-2 semi-private
Modelle / Anthropic
Claude Opus · Release 2026-04-16
Anthropics Flaggschiff April 2026. 13 % Coding-Steigerung gegenüber Opus 4.6, 3× mehr Produktionsaufgaben gelöst, hochauflösendes Vision bis 3,75 MP, neuer Tokenizer. $5/$25 pro 1M Tokens, 1M-Kontextfenster.
Fähigstes allgemein verfügbares Claude-Modell. Ideal für komplexe Reasoning-, lang laufende Agent-Aufgaben und Code-Review mit hohem Einsatz.
80average across 5 tested families
Wo Claude Opus 4.7 auf jeder öffentlichen Benchmark-Quelle mit Zeile steht. Rang zählt jedes Modell mit neuester Zeile im selben Test — kein universeller Qualitätsscore.
| Benchmark | Family | Rang | Score | Quelle | Datum |
|---|---|---|---|---|---|
| Vending-Bench 2 | Agentic | #1/ 5 | Andon Labs Vending-Bench 2 | 2026-06-01 | |
| SWE-bench Verified | Coding | #3/ 23 | 88 | SWE-bench official leaderboard | 2026-05-30 |
| ARC-AGI | Reasoning | #5/ 14 | 76 | ARC Prize leaderboard | 2026-05-15 |
| LMArena (Chatbot Arena) | Human preference | #6/ 13 | 82 | LMArena leaderboard | 2026-05-25 |
| Vals Index | Agentic | #7/ 20 | 66 | Vals AI — Vals Index | 2026-06-04 |
| Artificial Analysis Intelligence Index | Reasoning | #8/ 67 | 54 | Artificial Analysis | 2026-07-21 |
| SciCode | Coding | #10/ 65 | 55 | Artificial Analysis | 2026-07-21 |
| GPQA Diamond | Reasoning | #15/ 69 | 91 | Artificial Analysis | 2026-07-21 |
| AA time to first token | Performance | #16/ 69 | Artificial Analysis | 2026-07-21 | |
| Humanity's Last Exam | Reasoning | #18/ 65 | 40 | Artificial Analysis | 2026-07-21 |
| AA-LCR | Long context | #21/ 64 | 70 | Artificial Analysis | 2026-07-21 |
| Terminal-Bench | Agentic | #23/ 69 | 69 | Terminal-Bench 2.0 leaderboard | 2026-05-20 |
| τ³-Bench Banking | Agentic | #28/ 64 | 29 | Artificial Analysis | 2026-07-21 |
| IFBench | Reasoning | #42/ 57 | 59 | Artificial Analysis | 2026-07-21 |
| AA output speed | Performance | #46/ 69 | Artificial Analysis | 2026-07-21 |
Die drei höchstscorierenden kartierten Modelle pro Capability-Family. Wo Claude Opus 4.7 auftaucht, ist es hervorgehoben.
Benchmark-Zeilen im öffentlichen Ledger für Claude Opus 4.7 in den letzten 120 Tagen. Ältere Zeilen stehen in der vollen Tabelle unten.
| Benchmark | Family | Score | Quelle | Tage her |
|---|---|---|---|---|
| AA-LCR | Long context | 70.33% | Artificial Analysis | 2 |
| AA output speed | Performance | 61 t/s | Artificial Analysis | 2 |
| AA time to first token | Performance | 16.72s | Artificial Analysis | 2 |
| Artificial Analysis Intelligence Index | Reasoning | 53.5% | Artificial Analysis | 2 |
| GPQA Diamond | Reasoning | 91.4% | Artificial Analysis | 2 |
| Humanity's Last Exam | Reasoning | 39.6% | Artificial Analysis | 2 |
| IFBench | Reasoning | 58.64% | Artificial Analysis | 2 |
| SciCode | Coding | 54.5% | Artificial Analysis | 2 |
| τ³-Bench Banking | Agentic | 28.87% | Artificial Analysis | 2 |
| Vals Index | Agentic | 66.1% | Vals AI — Vals Index | 49 |
| Vending-Bench 2 | Agentic | 10936.76 | Andon Labs Vending-Bench 2 | 52 |
| SWE-bench Verified | Coding | 87.6% | SWE-bench official leaderboard | 54 |
| LMArena (Chatbot Arena) | Human preference | 1412 | LMArena leaderboard | 59 |
| Terminal-Bench | Agentic | 69.4% | Terminal-Bench 2.0 leaderboard | 64 |
| ARC-AGI | Reasoning | 75.8% | ARC Prize leaderboard | 69 |
Ein hoher Composite, der eine schwache Family versteckt, ist eine Falle. Diese Balken zeigen Families ohne öffentlichen Test — und wo das Modell führt.
Hover or tab any dot for name, score, and input price.
Jeder Katalog-Benchmark für Claude Opus 4.7. Scores verlinken zur Originalquelle; Lücken heißen: noch keine öffentliche Zeile.
15 of 32 catalog benchmarks have a sourced row for Claude Opus 4.7.
75.8%
ARC-AGI-2 semi-private
53.5%
AA Intelligence Index v4.1
91.4%
GPQA Diamond (AA run)
39.6%
HLE (AA run)
58.64%
IFBench (AA run)
54.5%
SciCode (AA run)
87.6%
SWE-bench Verified
69.4%
Terminal-Bench 2.0 (audited harness)
66.1%
Vals Index
10936.76
Final bank balance (USD)
28.87%
τ³-Bench Banking (AA run)
70.33%
AA-LCR (AA run)
61 t/s
Median output tokens/s (1k prompt, default provider)
16.72s
Median time to first token (1k prompt, default provider)
1412
LMArena Elo (Style Controlled)