News · Microsoft · 2 min read
Event date: · Published:
Azure Content Understanding adds GPT-5.6, GPT-6 and GPT-6.1 models
Microsoft's Azure AI Foundry blog announced that Content Understanding now supports GPT-5.6, GPT-6 and GPT-6.1 models for document, speech and video workflows, publishing benchmarks to guide model choice by cost and quality.
Microsoft announced on its Azure AI Foundry blog on November 11, 2026, that Content Understanding now supports the GPT-5.6, GPT-6, and GPT-6.1 model families for document, text, speech, and video workflows, publishing benchmark numbers to help developers choose the right model for their workload.
What changed
- Content Understanding now supports the GPT-5.6, GPT-6, and GPT-6.1 model families alongside the existing GPT-5 series.
- GPT-6.1 Sol is the recommended starting point for standard document extraction.
- GPT-6 Luna is recommended when token cost is the top priority; with 5 labeled examples its accuracy rises from 91.82% to 94.16%.
- GPT-6 Astra is recommended for a quality premium, priced roughly 5 times Sol's completion-token cost.
- Published benchmarks: document extraction accuracy (schema-only) is 95.74% for GPT-6.1 Sol versus 95.95% for GPT-6 Astra, and document segmentation Segment F1 is 98.00% for GPT-6 Astra versus 97.10% for GPT-6.1 Sol.
- Video benchmarks: headline-generation F1 for GPT-6 Sol is 86.40% (plus or minus 0.70 percentage points), and short-video ad performance for GPT-6.1 Sol is 93.60% (plus or minus 0.40 percentage points).
What it means for builders
Publishing head-to-head accuracy and cost benchmarks across a widened GPT-6 lineup gives developers building document and video extraction pipelines on Azure concrete data to pick a model instead of guessing by price tier. Teams can weigh GPT-6.1 Sol's balance of cost and accuracy against GPT-6 Luna's lower token cost or GPT-6 Astra's quality premium, and use the segmentation and video benchmarks to match a model to a specific extraction task.
Availability
Microsoft notes that model availability is region-specific and must be confirmed for each target region before deployment.
Sources
- Content Understanding adds GPT-5.6, GPT-6, and GPT-6.1: Choose the right model for your workload (Microsoft Foundry Blog, November 11, 2026, official)
Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.