59.9%
AA Intelligence Index v4.1
Modelle / Anthropic
Claude Fable · Release 2026-06-09
Anthropics Mythos-class-GA-Release Juni 2026. #1 im Vals-Index (75,15 %) und Vals-Multimodal-Index (74,15 %). 1M-Kontext, 128K-Output, Safety-Klassifikatoren leiten Hochrisiko-Anfragen an Opus 4.8. $10 / $50 pro 1M Tokens.
Langfristige Agent-Arbeit — mehrtägige Coding-Migrationen, Finance-Analyse, vision-lastige Rebuilds und Research-Pipelines, die Sonnet- oder Opus-Sessions überdauern.
83average across 4 tested families
Wo Claude Fable 5 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 |
|---|---|---|---|---|---|
| Artificial Analysis Intelligence Index | Reasoning | #1/ 67 | 60 | Artificial Analysis | 2026-07-21 |
| DeepSWE | Coding | #1/ 6 | 70 | DeepSWE leaderboard v1.1 | 2026-06-20 |
| GDPval | Agentic | #1/ 6 | 100 | Artificial Analysis — GDPval-AA v2 | 2026-06-17 |
| Humanity's Last Exam | Reasoning | #1/ 65 | 53 | Artificial Analysis | 2026-07-21 |
| SciCode | Coding | #1/ 65 | 60 | Artificial Analysis | 2026-07-21 |
| Vals Index | Agentic | #1/ 20 | 75 | Vals AI — Vals Index | 2026-07-16 |
| Terminal-Bench | Agentic | #4/ 69 | 85 | Artificial Analysis | 2026-07-21 |
| Vending-Bench 2 | Agentic | #5/ 5 | Andon Labs Vending-Bench 2 | 2026-06-01 | |
| AA time to first token | Performance | #7/ 69 | Artificial Analysis | 2026-07-21 | |
| GPQA Diamond | Reasoning | #8/ 69 | 93 | Artificial Analysis | 2026-07-21 |
| τ³-Bench Banking | Agentic | #15/ 64 | 62 | Artificial Analysis — τ³-Bench Banking | 2026-06-17 |
| AA-LCR | Long context | #22/ 64 | 70 | Artificial Analysis | 2026-07-21 |
| IFBench | Reasoning | #39/ 57 | 63 | Artificial Analysis | 2026-07-21 |
| AA output speed | Performance | #40/ 69 | Artificial Analysis | 2026-07-21 |
Die drei höchstscorierenden kartierten Modelle pro Capability-Family. Wo Claude Fable 5 auftaucht, ist es hervorgehoben.
Benchmark-Zeilen im öffentlichen Ledger für Claude Fable 5 in den letzten 120 Tagen. Ältere Zeilen stehen in der vollen Tabelle unten.
| Benchmark | Family | Score | Quelle | Tage her |
|---|---|---|---|---|
| AA-LCR | Long context | 70% | Artificial Analysis | 2 |
| AA output speed | Performance | 70 t/s | Artificial Analysis | 2 |
| AA time to first token | Performance | 68.33s | Artificial Analysis | 2 |
| Artificial Analysis Intelligence Index | Reasoning | 59.9% | Artificial Analysis | 2 |
| GPQA Diamond | Reasoning | 92.6% | Artificial Analysis | 2 |
| Humanity's Last Exam | Reasoning | 53.3% | Artificial Analysis | 2 |
| IFBench | Reasoning | 63.47% | Artificial Analysis | 2 |
| SciCode | Coding | 60.2% | Artificial Analysis | 2 |
| Terminal-Bench | Agentic | 84.64% | Artificial Analysis | 2 |
| Vals Index | Agentic | 75.1% | Vals AI — Vals Index | 7 |
| DeepSWE | Coding | 70% | DeepSWE leaderboard v1.1 | 33 |
| GDPval | Agentic | 1818 | Artificial Analysis — GDPval-AA v2 | 36 |
| τ³-Bench Banking | Agentic | 62.4% | Artificial Analysis — τ³-Bench Banking | 36 |
| Vending-Bench 2 | Agentic | 5680.26 | Andon Labs Vending-Bench 2 | 52 |
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 Fable 5. Scores verlinken zur Originalquelle; Lücken heißen: noch keine öffentliche Zeile.
14 of 32 catalog benchmarks have a sourced row for Claude Fable 5.
59.9%
AA Intelligence Index v4.1
92.6%
GPQA Diamond (AA run)
53.3%
HLE (AA run)
63.47%
IFBench (AA run)
1818
GDPval-AA v2
84.64%
Terminal-Bench v2.1 (AA run)
75.1%
Vals Index
5680.26
Final bank balance (USD)
62.4%
τ³-Bench Banking pass@1
70%
AA-LCR (AA run)
70 t/s
Median output tokens/s (1k prompt, default provider)
68.33s
Median time to first token (1k prompt, default provider)