Nearmap AI Gen 6 vs Claude Fable and Gemini
In the 72 hours that Claude Fable was available, we benchmarked it against Nearmap AI. Here are the results.
Read the full report →
As general-purpose LLMs have matured, a fair question has emerged: do purpose-built AI models still hold a measurable advantage for defined property tasks? In June 2026, Nearmap ran a controlled benchmark comparing Nearmap AI Gen 6 against seven generalist models from Google and Anthropic, including Claude Fable 5. The test was run across four property intelligence tasks, including pool detection, roof area, roof count and roof condition.
)
)
The benchmark
0
0.0
%
0
+
0
)
AI Accuracy
The results
Findings indicate a material performance gap across accuracy, failure mode distribution, and operational throughput, with structural causes that are unlikely to be resolved by improvements in general model capability alone.