Model Releases
we built parsebench because doc parsing didn’t have a benchmark people trusted @JensenHuang used it to rank NVIDIA’s model on stage today 🫡…
we built parsebench because doc parsing didn’t have a benchmark people trusted @JensenHuang used it to rank NVIDIA’s model on stage today 🫡 just another tuesday at @llama_index 🦙 Congrats to @nvidia f
we built parsebench because doc parsing didn’t have a benchmark people trusted @JensenHuang used it to rank NVIDIA’s model on stage today 🫡 just another tuesday at @llama_index 🦙 Congrats to @nvidia for creating the best open-weight model on document parsing so far. Nemotron 3.5 Parse takes the #1 spot on ParseBench, our comprehensive real-world benchmark for document understanding. The full list consists of 150+ frontier models, open-weight models, and s…
Source: Jerry Liu (X) | 2026-10-21