AI certifications, decoded
73 AI credentials from 32 providers, compared on the same scale. Find out what each one tests, how rigorously, and whether it is still current.
What does this certification prove?
Compare credentials on practical assessment, identity assurance, depth and verifiability, and learn the naming traps to watch for on a CV.
Read the hiring guide → For developersWhich certification should I take?
See what each exam covers, from generative AI apps and agents to MLOps, with cost, duration and validity side by side.
Compare certifications →How the comparisons work
Every credential is rated from 0 to 4 on the same rubric and shown as a Harvey ball. Coverage ratings follow each provider’s published exam weights. The methodology explains each scale and its limits.
What changed in 2026
- Microsoft retired AI-102 and DP-100 and launched a new line-up led by AI-103.
- AWS retired its ML Specialty, made the Generative AI Developer Professional generally available and began replacing the ML Engineer Associate exam.
- Agent-focused exams arrived from Google Cloud, Databricks, the Linux Foundation and Anthropic.
Read the full summary of 2026 changes
Guides
Certificate or certification: what is the difference?
The two words look alike on a CV and mean different things. How to tell them apart, and where microcredentials fit.
AI certifications: what changed in 2026
Retirements, rewrites and new agent-focused exams. A summary of a year in which most major providers reshaped their AI certifications.
How to choose an AI certification as a developer
A practical way to narrow down the options based on the platform you use, your experience and what you want the credential to do for you.
Are AI certifications worth it for developers?
What a certification can and cannot do for your career, and how to get the most out of the time you put in.
51 certifications can be taken today; 7 are retired or retiring but still appear on CVs. Researched on 2026-09-27. AI Developer Certificate is independent of every provider.