GPT-5.6 Beats Claude Mythos and Fable 5 in Benchmarks
The GPT-5.6 Soul Ultra model outperformed Claude Mythos on Stanford's terminal benchmark and beat Claude Fable 5 by a significant margin. A separate model, Terra, surpassed Fable 5 in cybersecurity tests.
The GPT-5.6 Soul Ultra model outperformed Claude Mythos on Stanford's terminal benchmark.
It also beat Claude Fable 5 by a significant margin in the same testing.
In addition, the Terra model surpassed Fable 5 in cybersecurity tests.