Across 19 shared benchmarks, Claude Opus 4.8 scores higher on 18 and Nemotron 3 Nano Omni on 1. The widest gap is Terminal-Bench Hard, where Claude Opus 4.8 scores 58.3 against 8.3. Nemotron 3 Nano Omni is the cheaper of the two on tracked API pricing ($0.07 against $5.00 per million input tokens).
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (anthropic-official, direct), otherwise the lowest tracked offer.