Independent evaluation platform
Artificial Analysis
A broad view of model capability, with the evaluation recipe made explicit.
Independent sources. A clearer perspective.
AI rankings & benchmarks.
Different tests answer different questions.
Find the evaluations that help you understand what a model can do.
An editorial guide, not a combined ranking. Scores from different methods are not directly comparable.
11 sources to explore
Methods before conclusionsHuman preference leaderboard
Arena
What people prefer when two answers are put side by side.
Technical benchmark suite
LiveBench
Release-based technical evaluation, with a newer long-horizon branch.
Professional-task evaluation platform
Vals AI
A view of AI performance grounded in particular kinds of professional work.
Generalisation benchmark family
ARC Prize
Can a system work out a new rule, rather than recognise a familiar answer?
Terminal-agent benchmark
Terminal-Bench
Whether an agent can complete work through a real terminal environment.
Software-engineering benchmark
SWE-bench Pro
Repository-level issues that require more than generating a short function.
Conversational-agent benchmark
τ³-bench
Following policy, using tools and retrieving knowledge through a conversation.
Tool-use benchmark
Berkeley Function Calling Leaderboard
Selecting tools and supplying the right arguments, across increasingly involved interactions.
Task-horizon research
METR
How difficult a task can an agent complete, expressed in human task time?
Evaluation platform and leaderboard family
OpenCompass
A broad evaluation ecosystem whose different leaderboards need their own context.