GPT-5.2 in Microsoft Foundry: the new standard for enterprise AI
The age of AI small talk is over. Teams need a reliable reasoning partner that plans, decides, and ships with confidence.
GPT-5.2 and GPT-5.2-Chat are now generally available in Microsoft Foundry. Built for developers and technical leaders, this series brings deeper reasoning, agentic execution, and compliance-ready controls on an enterprise platform.
Why GPT-5.2 matters now
Enterprise work isn't a clever chat. It's planning multi-agent workflows, turning briefs into code and tests, and shipping outcomes you can audit.
GPT-5.2 meets that bar: stronger logic, richer context handling, and end-to-end coordination-backed by Foundry's identity, policy, and governance layers.
GPT-5.1 vs. GPT-5.2: what changed
- Deeper logical chains: Breaks down complex tasks, justifies choices, and outputs explainable plans.
- Richer context: Ingests larger inputs (briefs, codebases, notes) and keeps the thread across long sessions.
- Agentic execution: Produces shippable artifacts with fewer loops-design docs, runnable code, unit tests, and deployment scripts.
- New architecture: Higher performance, better efficiency, and more reasoning depth compared to prior generations.
- Safety upgrades: Trained on the proven GPT-5.1 dataset with improved safeguards and integrations.
Result: fewer back-and-forths, more high-quality outputs you can hand to your pipeline.
Model lineup
- GPT-5.2: Advanced reasoning for hard problems with cleaner output. Great for information work-structured analysis, polished communication, and better formatting across docs, spreadsheets, and slides.
- GPT-5.2-Chat: Efficient for daily work and learning. Strong performance on info-seeking, how-to guides, technical writing, translation, study support, and practical career guidance.
Built for enterprise agents
- Multi-step logic: Plans, tests assumptions, and explains decisions.
- Context-aware planning: Reads large inputs for holistic, actionable output.
- Agentic execution: Coordinates design → implementation → testing → deployment to reduce manual oversight.
- Safety and governance: Managed identities and policy enforcement for secure, compliant adoption.
For teams building accountable systems, align guardrails with frameworks like the NIST AI Risk Management Framework.
Where teams will see impact
- Analytics and decision support: "Wind tunnel" strategies, explain trade-offs, and produce defensible plans for stakeholders.
- Application modernization: Refactor services, generate tests, and create migration plans with risk and rollback criteria.
- Data and pipelines: Audit ETL, recommend monitors/SLAs, and generate validation SQL for data integrity.
- Customer experiences: Build context-aware assistants and agent workflows that slot into existing apps.
The outcome: agents that stay reliable through complex workflows and service handoffs, while producing structured, auditable outputs that scale in Foundry.
Deployment and pricing
- GPT-5.2 - Standard Global: Input $1.75, Cached Input $0.175, Output $14.00 (USD per million tokens)
- GPT-5.2 - Standard Data Zones (US): Input $1.925, Cached Input $0.193, Output $15.40 (USD per million tokens)
- GPT-5.2-Chat - Standard Global: Input $1.75, Cached Input $0.175, Output $14.00 (USD per million tokens)
Use cached input pricing for repeat prompts and shared system instructions to lower unit costs.
Start building with GPT-5.2 today
Explore GPT-5.2 in Foundry today. Build where enterprise agents go from vision to production.
- Pick the fit: GPT-5.2 for complex reasoning; GPT-5.2-Chat for daily workflows.
- Scope your context: connect briefs, repos, knowledge bases, and policies.
- Wire tools and permissions: repos, CI/CD, data stores, ticketing, observability.
- Enforce guardrails: managed identities, policy rules, and logging.
- Ship a thin slice: design doc → code → unit tests → deployment script → review.
- Measure: cost, latency, quality, safety, and iteration count. Tune prompts, caching, and tool strategies.
If you're upskilling teams for agentic systems, browse practical learning paths by role at Complete AI Training - Courses by Job.
Your membership also unlocks:
