Artificial Analysis Launches Optima for Custom LLM Benchmarking

ArtificialAnlys · x · 2026-08-13

Leading AI evaluation platform Artificial Analysis has announced the launch of Optima, a new platform designed to help developers and enterprises build custom benchmarks tailored to their specific business use cases.

Optima integrates Artificial Analysis’ deep research expertise. It offers three ways to build a benchmark: uploading local datasets, importing from Hugging Face, or importing agent traces from various platforms. The platform's agent can automatically draft tasks and grading rubrics.

Optima supports various evaluation types including objective Q&A, document input, agentic tool usage, and interaction simulations. It then evaluates models on performance, time efficiency, and cost, helping users find the best model or an equally performant alternative at up to 10x lower cost.

Related event: Artificial Analysis Launches Optima for Custom LLM Benchmarks(2 posts)→

Original post →

More from Apps

Apps channel →