Claude Opus 5 runs on all three major clouds as well as Anthropic's own API. For enterprises, that usually settles the question — you buy where your data residency, committed spend, and procurement rules already point. What varies is which features come with you.
Model IDs by platform
| Platform | Model ID | Auth |
|---|---|---|
| Claude API | claude-opus-5 | Anthropic API key |
| Amazon Bedrock | anthropic.claude-opus-5 | AWS credentials |
| Google Cloud Vertex AI | claude-opus-5 | Google Cloud ADC |
| Microsoft Foundry | claude-opus-5 | Foundry credentials |
Only Bedrock adds a provider prefix. Sending a bare claude-opus-5 to Bedrock, or a prefixed ID to any other platform, returns an error.
What you give up on the cloud platforms
The core Messages API — streaming, tool use, adaptive thinking, effort levels, prompt caching, the 1M context window, vision, PDF input — works everywhere. The gaps are at the edges:
| Feature | Claude API | Bedrock | Vertex AI | Foundry |
|---|---|---|---|---|
| Fast mode | Yes | No | No | No |
| Message Batches | Yes | No | No | No |
| Web search tool | Yes | No | Yes | Beta |
| Code execution | Yes | No | No | Beta |
| Files API | Beta | No | No | Beta |
If your architecture depends on server-side tools, check this before committing — it is the most common source of an unpleasant surprise late in a cloud migration.
Choosing a platform
- Claude API — the fullest feature set and the earliest access to new capabilities.
- Amazon Bedrock — AWS-native IAM, billing, and data residency, at the cost of some server-side tooling.
- Google Cloud Vertex AI — the natural home if you are standardised on Google Cloud, with web search available.
- Microsoft Foundry — Azure-native, billed through the Microsoft Marketplace at standard API rates.
More on Claude Opus 5
Start with the complete Claude Opus 5 guide for the overview, or go deeper:
Ready to go deeper?
Read the full Claude Opus 5 guideFrequently Asked Questions
Is Claude Opus 5 available on AWS?
Yes, on Amazon Bedrock as `anthropic.claude-opus-5` — note the provider prefix, which the other platforms do not use. Core Messages API features work there, but fast mode, Message Batches, and the server-side web search and code execution tools are not available on Bedrock.
Does the model behave differently on each cloud?
The model is the same; the surrounding platform features differ. Adaptive thinking, effort levels, prompt caching, the 1M context window, and vision work everywhere. Fast mode is Claude API only, and server-side tool availability varies by platform.
Which platform should an enterprise pick?
Usually whichever cloud you already buy from, since data residency, committed spend, and procurement normally decide it. The exception is a workload that depends on features only the Claude API offers — fast mode, batches, or the full server-side tool set.
