readable AI benchmarks (simplified)

readable AI benchmarks

Anthropic

Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)

reasoning · closed weights · Sep 1, 2026

Claude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback) has a Quality Score of 74.3, ranking 9th among 272 scored models. Per 1M tokens, pricing is $10.00 input, $50.00 output, $0.250 cached input, and $12.50 cached output. It accepts text and image input, outputs text, and has a 1M token context window. Its Reliability Score is 67.1, ranking 8th among 272 scored models, above average (total average: 34.1). Its Value Score is 42.1, high compared with the total Value average of 28.2.

Quality Score74.3coverage 91.2%: missing Epoch General ECI, CursorBench 3.2 and DeepSWE v1.1
Value Score42.1
Reliability67.1
Cache Discount98%

Model specification

Reasoning
reasoning
Input modalities
text, image
Output modalities
text
Context window
1M tokens
Weights
closed weights

Token prices USD per 1M tokens

Input
$10.00
Output
$50.00
Cached Input
$0.250
Cached Output
$12.50
Cost per task
$0.774

Capability

Intelligence
58.1
Coding
75.2
Agentic
49.2
Omniscience
34.1
Correct
60.2%
Blended price
$7.18

Answer outcomes

Attempt rate: 86.3%

Correct 60.2%Incorrect 26.1%Abstained 13.7%

Artificial Analysis benchmarks

GDPval-AA v2
54%
τ³-Banking
39%
Terminal-Bench v2.1
85%
SciCode
56%
Humanity’s Last Exam
49%
GPQA Diamond
88%
CritPt
28%
AA-Omniscience
67%
AA-LCR
79%

Similar models