Amazon Nova foundation model
Amazon Nova Premier
Most capable Nova understanding model with the largest context window for complex reasoning and agentic tasks.
Key points
- Accepts text, image, and video input and outputs text.
- Is the most capable Nova understanding model.
- Supports about a 1M token context window.
- Can be used as a teacher model to distill smaller, faster custom models.
- Supports more than 200 languages, with optimization for a core language set.
When to use it
- Choose Premier for the most complex reasoning and multi-step agentic tasks.
- Use it when the largest Nova context window is required.
- Use it as a teacher model when distilling smaller custom models.
Exam tips
- Premier is the most capable Nova understanding model and has about a 1M token context window.
- Premier is multimodal for input but still outputs text.
- Do not choose Premier solely for lowest cost or lowest latency; Micro is the fastest and cheapest.