Flagship Nemotron model for high-throughput reasoning and complex agents
Details
| Field | Value |
|---|---|
| id | nvidia/Llama-3_1-Nemotron-Ultra-253B-v1 |
| family | nemotron |
| modality | text->text |
| context_length | 128000 |
| max_input_tokens | 120000 |
| max_output_tokens | 4096 |
| quantization | fp8 |
| knowledge_cutoff | 2024-12 |
| release_date | 2025-01-15 |
| is_open_weights | yes |
| supports_tool_call | yes |
| supports_reasoning | no |
| supports_structured_output | yes |
| supports_attachment | no |
| input_modalities | text |
| output_modalities | text |