Artificial Analysis

Description: Independent performance analysis of 100+ LLMs with detailed metrics

Website: https://artificialanalysis.ai

Artificial Analysis offers detailed, independent performance comparisons of over 100 LLM models. The platform measures intelligence, speed, price and latency on dedicated hardware.

Metrics

• Intelligence Index v4.0: Aggregates 10 independent evaluations (GDPval-AA, SciCode, GPQA Diamond, etc.)
• Output Speed: Tokens per second (e.g. Gemini 2.5 Flash-Lite: 439 t/s)
• Latency: Time to first token (e.g. DeepSeek-OCR: 0.19s)
• Price: Cost per million tokens
• Context Window: Supported token length

Top Performers

• Intelligence: GPT-5.2, Claude Opus 4.5
• Speed: Gemini 2.5 Flash-Lite, Nova Micro
• Price: Gemma 3n E4B ($0.03/M), DeepSeek-OCR ($0.05/M)
• Context: Llama 4 Scout (10M tokens)

Use Case

Ideal for objective comparisons when selecting models based on concrete performance data.