The AI Model Tier List
Which AI model is actually the best right now? This tier list answers that question and keeps answering it: the rankings below are rebuilt automatically every week from public leaderboard data (LMArena user-preference Elo, benchmark indexes) plus our own hands-on testing. Bookmark it; the answer changes more often than you would think.
Start Aura Farming
Top AI money moves delivered every morning - free forever.
Want to Build a Site Like This One?
The AI Money Farm is the exact step-by-step blueprint behind AIAuraFarm.com.
Get It on Amazon →The Current Top of the Arena
How This Tier List Works
Most tier lists are one person's opinion frozen in time. This one is a pipeline. Every week, an automated job pulls current standings from public leaderboards, LMArena's user-preference Elo ratings, the Artificial Analysis Intelligence Index, and published coding benchmarks, blends them with our own hands-on testing notes, and rewrites the data file this page reads. When the rankings shift, this page shifts with them, no human procrastination involved.
What the tiers mean in practice: S tier models are the ones you reach for when quality matters more than cost. A tier models are within striking distance and often the smarter buy. B tier is where value lives: open-weight and small models that punch far above their price. C tier is legacy technology that still works but no longer earns new projects. D tier is a museum.
One honest caveat: at the very top, the gap between S-tier models is now smaller than the gap between tasks. The best model for agentic coding, the best for multimodal analysis, and the best for creative writing are frequently three different names. Use the "best for" notes on each chip, and see our head-to-head breakdowns for the details: ChatGPT vs Claude, Claude vs Gemini, and ChatGPT vs Gemini.
Which Tier Should You Actually Pay For?
If you are earning money with AI, the arithmetic is simple. An S-tier subscription around $20 a month pays for itself if it saves you one hour. API users should route by task: S-tier models for the 10 percent of calls where quality is the product, A and B tier for the 90 percent that are summarization, extraction, and glue work. Our AI cost calculator does this math for your actual usage numbers.
For a deeper look at what each frontier lab is shipping and how to get started with each tool, see the getting-started guides in our resources hub.
Frequently Asked Questions
How often does this page update?
The data file behind it is rebuilt weekly by an automated job, and the page renders whatever is current every time you load it. The "updated" badge at the top shows the date of the last refresh.
Why does your ranking differ from raw LMArena?
Raw arena Elo measures which answers people prefer in blind votes, which rewards style as well as substance. We blend it with benchmark indexes and practical testing, so a model that charms voters but fumbles real work lands lower here.
Where are the image and video models?
This list covers text and reasoning models. Image, video, and audio tools have their own dynamics; our AI directory covers those categories.
