Generative AI platforms & models · AWS vs Google Cloud
Amazon Nova vs Gemini models on Agent Platform
Amazon Nova's Google Cloud counterpart is Google's own model family on Agent Platform: Gemini for text, multimodal, and reasoning, plus Veo for video, Lyria for music, and Gemma open models.
Side-by-side comparison
| Aspect | Amazon Web Services | Google Cloud |
|---|---|---|
| Service | Amazon Nova | Gemini models on Agent Platform |
| Lifecycle | Generally available | Generally available |
| What it is | Amazon's first-party foundation models. Understanding (Nova Micro/Lite/Pro, Nova 2 Lite), creative (Nova Canvas image, Nova Reel video), speech-to-speech (Nova 2 Sonic), and multimodal embeddings. | Google's first-party models (Gemini, Gemma, Veo, Lyria, embeddings) served per token or per output unit through the Agent Platform API. |
| Pricing model | Nova is priced per token (text), per image, or per second of video on Bedrock, with batch and Provisioned Throughput options. | Gemini and partner models are billed per input/output token (media models per image, second, or character), with Standard, Priority, and Flex pay-as-you-go tiers and batch discounts. Prices vary by model and region. |
| Docs | Amazon Nova docs | Gemini models on Agent Platform docs |
| Pricing page | Amazon Nova pricing | Gemini models on Agent Platform pricing |
Key differences
- Gemini models are multimodal by default (text, image, audio, video, and documents in one request).
- Gemma is an open-weights family you can also self-host, which Nova has no direct counterpart for.
- Both vendors retire model versions on published schedules. Check the model lifecycle page before pinning a version.
Pricing summary
Pricing models only. No prices are quoted because they change by region and over time.
- AWS
- Nova is priced per token (text), per image, or per second of video on Bedrock, with batch and Provisioned Throughput options. Amazon Nova pricing →
- GCP
- Gemini and partner models are billed per input/output token (media models per image, second, or character), with Standard, Priority, and Flex pay-as-you-go tiers and batch discounts. Prices vary by model and region. Gemini models on Agent Platform pricing →
And on Azure?
Partial match Azure's equivalent is Microsoft models in Foundry (MAI, Phi, Model Router), Microsoft Foundry Models (incl. Azure OpenAI). AWS vs Azure comparison →
FAQ
What is the Google Cloud equivalent of Amazon Nova?
Amazon Nova's Google Cloud counterpart is Google's own model family on Agent Platform: Gemini for text, multimodal, and reasoning, plus Veo for video, Lyria for music, and Gemma open models.
How does pricing compare between Amazon Nova and Gemini models on Agent Platform?
AWS: Nova is priced per token (text), per image, or per second of video on Bedrock, with batch and Provisioned Throughput options. Google Cloud: Gemini and partner models are billed per input/output token (media models per image, second, or character), with Standard, Priority, and Flex pay-as-you-go tiers and batch discounts. Prices vary by model and region. Check the official pricing pages for current rates.
What is the Azure equivalent?
Microsoft models in Foundry (MAI, Phi, Model Router) and Microsoft Foundry Models (incl. Azure OpenAI) (partial match). Amazon Nova's closest Azure counterpart is Microsoft's own models in Foundry (MAI and Phi, plus Model Router).