Optima tackles AI benchmarking's biggest flaw by letting users test models against their own data

Artificial Analysis has launched Optima, a platform that lets users build custom AI benchmarks from their own data and workflows, comparing models on quality, cost, and time per task. The tool addresses a key flaw in AI benchmarking by enabling tests against user-specific data, which is particularly useful for agent-based applications where these metrics are more informative than raw token pricing.