1MContext
FreeIn, per 1M
FreeOut, per 1M
- Lab
- NVIDIA
- Released
- 4 Jun 2026
- Context window
- 1M tokens
- Longest answer
- 66K tokens
- Input, per 1M tokens
- Free
- Output, per 1M tokens
- Free
- Reads images
- No
- Reasoning
- Yes
- Tool calling
- Yes
- Open weights
- Yes
- Answers in
- text
- Model id
- nvidia/nemotron-3-ultra-550b-a55b:free