NVIDIA / nvidia/nemotron-3.5-super-vl

Nemotron 3.5 Super VL - access through LLMTR

Nemotron 3.5 Super VL is NVIDIA's open multimodal Mixture-of-Experts model, running 12 billion of its 120 billion parameters per token. It reads text, images and short videos and answers in text. It offers a 262,144-token context window, native tool calling with `auto`, `required` or a named function, JSON object mode and JSON Schema structured output. Send images as `image_url` and videos as `video_url` content parts carrying a base64 data URL. Step-by-step reasoning is on by default and returned separately from the answer, so even a short question spends reasoning tokens; switch it off per request when you need a quick answer. Reasoning tokens bill as output. `max_tokens` is accepted up to what the context window leaves after the prompt; check `finish_reason` on long answers.