Évaluation Updated 2026-04

AI Benchmark

Definition

An AI benchmark is a standardized test that measures and compares AI model performance on specific tasks.

See also in the glossary

L
LLM (Large Language Model)
An LLM is an AI model trained on billions of texts, capable of understanding and generating human language.
F
Foundation Model
A foundation model is a large AI model pre-trained on massive data, adaptable to multiple tasks.
A
AI Hallucination
An AI hallucination is a response generated by an AI model that appears plausible but is factually incorrect or fabricated.
A
AI Reasoning
AI reasoning refers to a model's ability to break down a problem into logical steps to reach a conclusion, rather than answering instinctively.

Tools that use ai benchmark

C
ChatGPT
The world's most used conversational AI assistant
4.6/5
C
Claude
The AI that understands nuance, by Anthropic
4.7/5
G
Gemini
Google's AI assistant with 1M token context
4.5/5
D
DeepSeek
The open source Chinese model rivaling GPT-4
4.7/5

Frequently Asked Questions

What are the most popular AI benchmarks?
MMLU (general knowledge), HumanEval (code), MATH (mathematics), HellaSwag (reasoning), and LMSYS Arena ELO (human voting ranking).
Are benchmarks reliable?
Partially. Models can be optimized to perform on benchmarks without being better in practice. Arena ELO ranking is considered the most representative.