CMU, NVIDIA and Meta Release Singularity-Bench; Opus 5.5 Beats Human Baseline
CMU, NVIDIA and Meta released Singularity-Bench, an open-ended benchmark testing whether AI agents can autonomously build core AI systems across six tasks. Opus 5.5 with Claude Code built AI infrastructure end-to-end, outperforming SGLang and exceeding human baselines with a Singularity Index of 1.25.
2026-11-16 ~ 2026-11-17 · 2 related posts