65 ms·
Benchmarks are far from everything, but I would love to see the outcome of an experiment benchmarking GPT-4o (which is one of the earlier models with a >100k co
by aesthesia 15d ago
Benchmarks are far from everything, but I would love to see the outcome of an experiment benchmarking GPT-4o (which is one of the earlier models with a >100k context window) against GPT-5.6 or Opus 5 in modern harnesses.