Azure OpenAI Fine-Tuning

source

closed

training-code

closed

runs-on

Azure-AI-Foundry-only

license

Proprietary(managed Azure service)

A proprietary managed fine-tuning service inside Microsoft Foundry, with no source published. It runs only on Azure, and tunes both closed OpenAI models, which are generally available, and open-weight models in preview: Ministral, Qwen, Llama and gpt-oss.

• https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/fine-tuning recorded 2026-09-23

  Supported-models table lists gpt-4o-mini as SFT-only; gpt-4o and the gpt-4.1/mini/nano trio as SFT+DPO; o4-mini and gpt-5 as RFT (gpt-5's RFT access gated by invitation); and public-preview SFT for Ministral-3B, Qwen-32B, Llama-3.3-70B-Instruct, and gpt-oss-20b. Fine-tuning runs only inside a Foundry project resource under the Foundry Owner/User RBAC roles; no repository or self-host path is offered anywhere on the page.

• https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy recorded 2026-09-23

  "Your fine-tuned Models sold by Azure are available exclusively for your use." "Foundry is an Azure service; Microsoft hosts the Models sold by Azure in Microsoft's Azure environment and Models sold by Azure do NOT interact with any services operated by providers of Models sold by Azure, for example, OpenAI." Fine-tuned models ("base or fine-tuned") are deployed in the customer's Foundry resource and pass through the same real-time content-filtering pipeline as base models.

• https://www.microsoft.com/licensing/terms/product/ForOnlineServices/all recorded 2026-09-23

  Defines "Covered Product" as "any Azure OpenAI model in Microsoft Foundry Models or any Copilot (excluding free Previews), in either case, that is available for a fee through Microsoft volume licensing or used with a paid subscription to an Online Service," and bars use of the Online Service "to reverse engineer any Online Service or exfiltrate the weights of any AI models or otherwise discover any underlying components, algorithms or systems included in the Online Service."

Microsoft publishes no usage figures for this service, such as jobs run or customers tuning models. Azure OpenAI is widely deployed, but nothing published is specific to fine-tuning, so the map leaves this product unscored rather than borrowing a figure from the umbrella service.

• https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/fine-tuning recorded 2026-09-28

  The article carries no usage figure; searched the body for customers, users, downloads, jobs run and adoption: the only hits are the sample training line "About 93 million miles. Just a quick drive, really." and the advice "hundreds, if not thousands, of training examples", neither a usage metric.

Azure OpenAI Fine-Tuning offers supervised and preference tuning on OpenAI models such as gpt-4o, reinforcement fine-tuning on o4-mini, and supervised tuning of open-weight models such as Llama and Qwen, but exposes only batch size, learning rate and epoch count as settings. That gives it managed breadth rather than the deeper control over parallelism and precision that a training framework offers.

• https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/fine-tuning recorded 2026-09-28

  Supported-models table lists gpt-4o with methods "SFT, DPO" and o4-mini with "RFT" (gpt-5 RFT "gated and available by invitation only"), plus SFT on four open-weight models. The only tunable hyperparameters are batch_size, learning_rate_multiplier and n_epochs; searched the body for parallelism, precision, GPU, fp16 and bf16: none found.