Anthropic

Claude Fable 5 2 (max)

Released 2026-11-09 · Context 1000K · Data snapshot 2026-11-22

Overall score

76.3

Ranked #1
Benchmark coverage
65.4%
17 third-party benchmarks
Context window
1000K
tokens
Input price
$8.00
Cache read $0.20
Output price
$40.00
per million tokens
OpenRouter usage
#37
713.8B tokens/day

Prices converted at snapshot-day rate (USD/CNY 6.6777)

Dimension performance

Reasoning84.2
Coding76.9
Knowledge75.6
Office67.7
Above averageBelow averageAnchored at 50 (reference models’ average); every 15 points ≈ one standard deviation

Benchmark details

Coding7 items
Vibe Code Bench93.77
LiveBench 2.0 · Browser86.9
Code Migration v1.183.27
LiveBench 2.0 · Long Agentic82.9
LiveBench 2.0 · Coding79.2
Terminal-Bench 4.065.152
ProgramBench19
Reasoning4 items
FrontierMath (Tiers 1-3)max95.1
Mystery Game Puzzlesmax90
LiveBench 2.0 · Reasoning88.8
Chess Puzzlesmax79
Office4 items
LiveBench 2.0 · Research94.5
LiveBench 2.0 · Design85.1
LiveBench 2.0 · Data Analysis81.5
Legal Research46.15
Knowledge2 items
SimpleQA Verifiedmax94
GPQA Diamondmax93.9

Multiple reasoning tiers of the same benchmark take the best value; raw scores have different scales and are not comparable across benchmarks.

Score trend

2026-11-102026-11-22
View daily data
2026-11-2276.3
2026-11-2176.0
2026-11-2076.0
2026-11-1976.3
2026-11-1876.3
2026-11-1776.3
2026-11-1676.3
2026-11-1576.3
2026-11-1476.3
2026-11-1376.3
2026-11-1276.3
2026-11-1176.3
2026-11-1076.3
Scores and prices come from third-party public data (Artificial Analysis / LMArena / OpenRouter). 50 is the reference models’ average; every 15 points ≈ one standard deviation. Prices are converted at snapshot-day rates and provided for reference only. Return to LLM Leaderboard.