GPT-5.5 on Azure | RAGtimeZ

GPT-5.5 Now Available on Microsoft Foundry: New Developments in Enterprise AI Platforms

According to the Microsoft Azure blog, OpenAI’s GPT-5.5 is now available on Microsoft Foundry. Unlike previous cloud-based AI services, this provides a new option for companies to run, adapt, and deploy AI models under their own conditions.

Microsoft Foundry is not just an AI service, but is positioned as a “full-stack AI app and agent factory”. Through an integrated platform from cloud to edge, it enables developers to not only use AI, but also to create with AI.

Practicality of Open-Source Model gpt-oss

OpenAI has announced the gpt-oss model series as its first open-weight release since GPT-2. gpt-oss-120b can be run on a single enterprise GPU, and gpt-oss-20b supports local environment execution.

These models are not just simplified versions, but are designed with real-world deployment in mind. They provide fast and practical performance, from large-scale inference in the cloud to agent processing at the edge.

The open-weight nature of these models makes fine-tuning, distillation, and optimization easier. With domain-specific co-pilot development, compression for offline inference, and local prototyping before full operation, Azure AI Foundry and Foundry Local provide a comprehensive set of tools. (Source: OpenAI’s open‑source model: gpt‑oss on Azure AI Foundry and Windows AI Foundry)

Integrated Development Experience with Responses API

The new Responses API in Azure OpenAI provides a unified experience that integrates chat completion and Assistants API functionality. This API enables the generation of stateful multi-turn responses and supports the computer-use-preview model.

As a practical example, the following Python snippet uses the Responses API:

import os
from openai import OpenAI
client = OpenAI(
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),
    base_url="https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/",
)
response = client.responses.create(
    model="gpt-4.1-nano", # Replace with your model deployment name
    input="This is a test.",
)
print(response.model_dump_json(indent=2))

This API has the advantage of allowing developers to build complex conversational applications with more concise code compared to traditional separate APIs. (Source: Use the Azure OpenAI Responses API - Microsoft Foundry)

Expansion of Inference Models and Function Calling Support

The GPT-5 series includes inference-specialized models, such as o3-mini, o1, and o1-mini, which are now available. These models are designed specifically for complex logical inference tasks.

Regarding Function Calling, parallel function calls are supported in each version of GPT-4.1 and GPT-5 series. Specifically, the following models are available:

gpt-4.1 (2025-04-14)
gpt-5 (2025-08-07)
gpt-5.2-chat (2025-12-11)
gpt-5.3-codex (2026-02-24)

This enables developers to efficiently implement integration with multiple external systems. (Source: How to use function calling with Azure OpenAI in Microsoft Foundry Models)

Summary

Using GPT-5.5 and gpt-oss models, companies can build and operate AI applications tailored to their own requirements in their own environment
By introducing the Responses API, the development time for multi-turn conversational applications, which previously required combining multiple APIs, can be significantly reduced
Fine-tuning can be performed using the open-weight gpt-oss model, and domain-specific co-pilots can be optimized with the company’s own data
By utilizing the GPT-5 series with Function Calling support, external system integration processes can be executed in parallel, improving the response speed of agent-type applications