Claude Opus 5 Is Now Available in Microsoft Foundry

Anthropic's Claude Opus 5 — the first Opus model in the Claude 5 generation — is now available directly in Microsoft Foundry. It joins Claude Sonnet and Haiku as part of the growing Claude model family on Foundry, giving Azure customers a frontier-grade option for the reasoning-heavy, long-running work that lighter models struggle to finish reliably.

Combined with Foundry's enterprise platform — evaluation tools, native security controls, governance, and scalable deployment — teams can build and operate the next generation of AI applications without leaving the Azure ecosystem they already manage identity, networking, and compliance through.

Most models are optimised for short, single-turn exchanges. Opus 5 is built for the opposite: multi-step work that unfolds over minutes or hours, where the model has to keep track of what it has already tried, what failed, and what to do next.

• Advanced planning and reasoning — Opus 5 breaks a large task into an ordered sequence of steps before acting, rather than reacting one prompt at a time.
• Long-running tasks — it sustains context and objective across extended sessions, instead of losing the thread after a few exchanges.
• Adaptive execution — when a planned step fails or an obstacle appears, it revises the plan and keeps going rather than stalling out.

The headline capability behind Opus 5 is agentic reliability — recovering from errors, coordinating multiple tools, and reaching an objective without a human re-prompting it at every turn.

• Reason across multiple steps — Opus 5 holds a chain of intermediate decisions in context rather than treating each step in isolation.
• Use tools and functions — it calls external tools, APIs, and functions as part of executing a plan, not just as a one-off action.
• Coordinate workflows — it can orchestrate work across multiple tools and subagents to complete a larger objective.
• Adapt as tasks evolve — when requirements shift mid-task, it updates its plan instead of continuing down a now-invalid path.

Claude Opus 5 deploys through the Foundry portal like any other model in the catalog: open Discover → Models, select Claude Opus 5, choose Deploy, accept the Azure Marketplace terms, and pick a region scope (Global or Data Zone, where available). You need Contributor or Owner on the resource group, and a paid Azure subscription with a supported billing region.

Calling the model

Once deployed, you call Opus 5 through the Claude Messages API using the Anthropic SDK, authenticating with either Microsoft Entra ID (keyless, via DefaultAzureCredential) or a deployment API key. The base URL follows the pattern https://<resource-name>.services.ai.azure.com/anthropic.

from anthropic import AnthropicFoundry
from azure.identity import DefaultAzureCredential, get_bearer_token_provider

baseURL = "https://<resource-name>.services.ai.azure.com/anthropic"
deploymentName = "claude-opus-5"  # the name you chose during deployment

# Keyless auth via Microsoft Entra ID
tokenProvider = get_bearer_token_provider(
    DefaultAzureCredential(), "https://ai.azure.com/.default"
)

client = AnthropicFoundry(
    azure_ad_token_provider=tokenProvider,
    base_url=baseURL
)

message = client.messages.create(
    model=deploymentName,
    messages=[
        { "role": "user", "content": "Plan and execute a 3-step migration checklist for moving a service from VMs to AKS." }
    ],
    max_tokens=2048,
    temperature=1,
    thinking={"type": "adaptive"},
    stream=False
)

print(message.content)

Prefer API key authentication for quick prototyping — swap the `azure_ad_token_provider` for `api_key=<your-key>`. In production, Entra ID is the safer default: no long-lived secret to rotate or leak, and access is governed through the same RBAC roles you already use for other Azure resources.

• More choice in Foundry — access Claude Opus 5 alongside other leading models in the same catalog, without standing up a separate vendor integration.
• Stronger agentic workloads — handle complex, multi-step tasks with greater reliability than lighter models allow.
• Higher performance — advanced reasoning and long-running execution for the demanding use cases that used to require a human in the loop at every step.
• Seamless ecosystem — native integration with the Microsoft tools, data, and governance your organisation already runs on.

For teams already standardised on Azure and Foundry, the practical effect is simple: the same identity, networking, monitoring, and compliance controls now extend to Anthropic's most capable model, with no new vendor relationship to manage.

Claude Opus 5 in Microsoft Foundry is aimed squarely at the workloads that make or break an agentic AI strategy: long-running, multi-step tasks where the model has to plan, adapt, and recover without constant supervision. Combined with Foundry's governance and deployment tooling, it gives Azure-native teams a frontier reasoning model without leaving their existing security and compliance boundary.

If you are already running workloads in Foundry, deploying Opus 5 for a pilot agentic use case — a migration assistant, a multi-step data pipeline, a codebase-navigation agent — is a low-friction way to see whether the extra reasoning depth is worth the move up from lighter models.