Amazon has launched its largest language model, the Amazon Nova Premier, which is larger than the previous Nova Pro model. It boasts a context window of 1 million tokens (compared to Pro’s 300,000 tokens) and can handle high complexity tasks, as well as be used as a parent model to teach other models through distillation.
Nova Premier surpasses Nova Pro in every benchmark, with Amazon claiming it to be the most advanced non-reasoning model in the industry. It can coordinate with agents and also has the advantage of being more cost-effective on Amazon Bedrock compared to similar models (input price of $0.0025 per 1,000 tokens, output price of $0.0125 per 1,000 tokens).
Source – AWS Blog
TLDR: Amazon introduces the powerful Amazon Nova Premier, a cutting-edge language model with enhanced capabilities and cost-efficiency.
Leave a Comment