AI Agent StoreNVIDIA · Download verified
Nemotron 3.5 Parse
Turn document images into readable text, tables and chart data while retaining reading order and locations on the page.
Released 2026-10-17 · First reviewed 2026-10-27 · Checked 2026-11-03.
We still need comparable independent tests. You can compare the confirmed features and limits below, but we cannot yet give this version an Agent Fit or Practical Value rating. Try it on your hardware and tasks before relying on it.
What you can use today
You download the model files and supply the hardware and software to run them. Capacity is the amount of information a request can contain; settings and available memory can limit what fits.
Exact download ID: nvidia/Nemotron-3.5-Parse. A specialized 631-million-parameter document parser. Its helpers return Markdown, structured blocks and raw output. Extra passes revisit tables and charts; test whether they recover useful detail in your documents.
Downloadable under OpenMDW 1.1 for commercial or non-commercial use. The reference software uses Linux with Transformers or vLLM; NVIDIA reports tests on H100 hardware.
The configuration has a 16,384-token sequence limit. Image processing and generated text share the model’s setup; this is not a general chat input allowance.
No hosted per-page or token price is established. The publisher’s accuracy and pages-per-second figures use its own testing setup and are not independent results here.
Provider release and specifications ↗
Check availability and plans ↗
Test it before switching
Give each model the same examples, tools and definition of success. Record correct outcomes, response time, retries and total cost. Include the time and hardware needed to run it.
Compare caching and long-context costs ↗